MOC System Missed Entry Orders

[JulianCohen]

That’s a good point. However if the system is part of a portfolio of systems then the leverage is not so much of a concern as the equity from the other systems effectively negates the leverage as far as IB sees it.

So far I have found that most of my returns come from the 1000 and if I run without that then I get low returns. If I include the 3000 to the 1000 it doesn’t beat my returns from the 1000 on it’s own.

[Anonymous]

Are you making other adjustments at the same time as widening the universe to the 3000?

For example perhaps on RUI you are at 10 USD minimum, 1 Mil ave vol minimum, 20 positions max. So when trying RUA perhaps you go to 5 USD min, 500K ave vol minimum, 50 positions max. Of course just examples, but it is only logical that the limits get moved when working on a broader universe that would have generally lower prices and volumes per symbol.

Trimming these things and ensuring you don’t exceed a reasonable daily trade size volume as percentage of entry bar volume (say, 5%, set this in the Portfolio tab in Backtester settings) could open you up to more trades whilst still working within reasonable limits of daily volume.

Trimming your percentage on the ATR stretch will give more signals and fills. Maybe you are running with 0.8 x ATR or even 1.0 x ATR so trimming this to say 0.6 or 0.5 ATR will give you more signals.

And as always, don’t get lost in the forest for the trees on the returns when comparing systems/results. Perhaps the new results you are seeing on the RUA are a lot more realistic and robust and is the more conservative result you should accept and trade. Perhaps your current system is fooling you into enjoying what looks like a nice juicy return, albeit with a bumpier ride along the way. E.g. Perhaps you need to accept that over a 10 year period the RUA system gives 14% Car on 16% drawdown across 10,000 trades with 4700 trades within -5% and 5100 trades within +5% and the remaining 200 trades in the +/- tails as compared to your current RUI system which is showing you 19% CAR on 17% drawdown across 2,000 trades with 400 in the -5% and 200 in the -10% and 150 in the -15% and 50 in the -20% and 50 in the remaining negative tails and 400 in the +5% and 250 in the +10% and 200 in the +15% and 150 in the +20% and 100 in the +25% and 50 in the remaining positive tails.

It is easy to chase a CAR/MDD dream but the proof is in the pictures regarding how well you are actually going to trade the thing in real time. Good looking equity curve is nice, but the profit distributions, monte carlo, underwater, monthly returns etc all need to actually make sense too. Don’t get lost in a dream in just the numbers on the Statistics page.

For me, I like lots of trades (my target is no less than 1000 per year on these reversion MOC systems), tall narrow profit distribution with more than 95% of the trades in just the +/-5% result and long tails making up no more than about 5% of the trades. And I know with ranking code Monte carlo means less, but on the monte carlo page I like to see the min/max equity curve with the best case green and worst case red scenario hugging the blue centre equity line like a rocket taking off into the sky with nice tight trail behind it. Some of my systems show this page like an old garbage truck chugging down the road with smoke spewing out the back left and right and it immediately tells me somethings wrong.

What columns/metrics are you looking at for your benchmarks, guides, optimisation etc? I’m guessing CAR/MDD, max DD etc, all the usual stuff? Are you optimising for anything in particular?

Also I always make sure that I use a very realistic number for the amount of cash I will likely end up using for a specific system when building/backtesting. If I know I am going to use around 200K USD and 2X leverage for this system in real trade, then I will set this as the parameters for backtesting and optimisation. Some people don’t do this and just use a generic 100K or $1mil because it is a very simple number to work from as a base and easy enough to see how much your system grows by and then even perhaps compare them to other systems. I see this as a danger though because your daily trade size limit and these issues as we are seeing here and now start to present themselves. Perhaps you built and backtested your system with 100K starting balance but in real trade you use 250K. if some of your positions are already at the limits of daily trade size volume at 100K then no wonder you can’t get a fill for the full amount you need now that you are trading it with 250K. Conversely perhaps you used 1 Mil as your backtesting base, but as the equity result starts to grow to 1.2 or 1.3 Mil you start to see daily trade size limit being hit and you abandon the system because you think it doesn’t work, however you would never have 1.3 Mil in trade anyway, or at least not for another 5 years if you are starting at 250 (and at which time you would probably have a different system or parameters anyway, so it is probably still an ok system for today and the coming years).

I haven’t had the “this security not available for leverage” issue. I purposefully leave a buffer of about 15% cash at all times to combat drawdown and margin calls and also have multiple systems across varying leverage that helps this issue.

[JulianCohen]

My RUI system runs off 0.4 ATR and 20 positions at 20% with volume at > 1,000,000

I tested with RUA running off 60 positions at 6.6% and reduced the volume to > 250,000 as that should cover the position size.

Dropping down to $5 min price adversely affected things for me, but I didn’t try $1 yet

I tend to optimize only the ATR stretch and I’m looking for the usual suspects, CAR MDD etc but also the win rate especially with a tight return system such as this.

I always test with the account size I plan to use.

I haven’t as yet looked into the systems to see what percentage of trades are in the long tail and that is my next stage to testing. So far I have found the best I can achieve is a drop in CAGR from 50% to 45% (not really substantial) and also a drop in MDD from 19% to 14% so it is looking quite good, but I’ll delve deeper into where the returns are coming from and let you know what I find.

It’s always an interesting experiment no matter what.

[TerryDunne]

Thanks for the really thoughtful replies. It’s becoming clear to me how much knowledge exists here and how generous people are in sharing it!

Said, I think you’re saying that you don’t have any slippage issues with your exits? That would be really encouraging if I’ve understood correctly as I’ve concluded that this is a much bigger risk than the entry side. I also now understand why testing with a minimum share price is important – selling at $1 but only getting 99c is much more of an issue than selling at $10 but only getting filled at $9.99.

Matt, my curiosity has really been piqued by the idea of trading (potentially) all stocks with a sufficient price and liquidity rather than only the ones in an index…can’t wait to get home to my special friend Amibroker! Are you putting on dozens of LIT trades each night?

One thing I haven’t understood is how my fills might improve using a dedicated server? Are you suggesting that speed will improve and that will lead to better fills?

Best wishes,

Terry

[Anonymous]

Julian do you have a regime/market trend filter? For me, I don’t run any filter on my MR MOC systems. I just run it every single day. Previous building/testing on only being in the market at certain times proved of no real benefit and only succeeded in reducing the number of trades I could get. For me and my system, the more trades and the more you are in the market is how it is showing success.

What else are you doing for deciding your entries? Any other indicators or metrics? How about ranking?

[Anonymous]

Terry, I do still trade within an index. For my MR MOC systems I use Russell 3000. Even the lowest value companies are (if my memory is right) something like 150 Mil USD market cap, so I’m still ok with that in terms of how large the companies are and investing some money in them for a few hours within any given day. At least they aren’t penny stocks lucky to trade 5,000 units in a day.

Yeah I usually put on lots of trades each night. For example last night I placed a total of 85 orders and got 19 fills (and lost money overall, a few more reds than greens, but that’s how it is. Every single day last week was a green day, so its swings and roundabouts…).

If the system is fully loaded it will be 120 orders and hopefully about 20 to 30 fills. Last Thursday was just such a day. It was so beautiful, 120 orders and I got 32 fills. Just as per my statistics about 45% of them were red and 55% of them green and made profit for the day. I just love it when the volume pumps through and you see your results come out perfectly as planned. Its easy to be on Nick’s journey of “next 1000 trades” when you have plenty of trades pumping through. I really am sleeping better now after placing the trades then going to bed.

Earlier this year when I was on the opposite end of this spectrum and depending more on the long tail trades to make my system profitable, I was placing 20 orders a day and in some weeks would get not one fill for the whole week. How useless did I feel when saying to WonderWife “sorry dear, no fills for the whole week”. Then in the next week I would get 10 fills producing nothing great positive nor negative.

I got so annoyed one day when I was out on the town for some reason and was not able to get back to my computer on time to place the orders and missed a Tesla fill were I would have made 34K in the one day. And sure, another time I missed an AMGN trade where I would have lost 15K. However it was from these lessons that I knew I had to dig deeper in my learning and work through trying to make greater volumes of trades work and reduce dependency on less frequent, large wins.

My downfall was falling in love with results in the statistics page like 35% CAR on 19% Drawdown and not giving enough attention to the total number of trades and the profit distribution curve. I was absolutely depending on the long tail trades. You should have seen the equity curve, edging up not too much for months in a row, bumping along, then a 2018 Feb or Oct 2018 comes along and a big spike upwards in the curve and 20 trades in the day. Here I was thinking “wow this is so perfect, it makes money when its bumpy! I can use my trend systems in all these quiet times, and then use these bumpy systems in all these volatile times and I have the perfect complementary system!!!”. It was far from the truth. It was impossible to trade having no results for months on end hoping and praying for the big when and then just as you guessed it, the day you miss is the day all the profit is made.

For the server thing…..

1. I do not use a VPS, I use a bare metal dedicated server which is allocated all to me, myself. I don’t really like VPS’s because frankly speaking you are only getting a virtual/time slice of the windows operating system on a server computer. For example perhaps your slice of resources might be 1/8th of the total CPU/ram/hard drive/ethernet resource available within that physical server. Understood why it is done, such that resellers can use bare metal servers to then re-sell 8 VPS products to customers, amortise the cost and make a few bucks. However if you are using a VPS from a provider that markets themselves to trading customers, then in my opinion it is highly likely that at 9.30am when the market opens there will be seven other trading customers with their systems/software running and demanding CPU/RAM/HDD/Ethernet time all at the very same time as when you critically need it. That cannot possibly be a good thing in my mind, especially if some greedy bugger is running very hungry software of some type on their 1/8th of the resource. It’s just impossible to believe something virtually split eight times on a physical resource can be the same as having that one resource all to yourself without sharing. Of course you shouldn’t then go for a VPS with a provider that DOESN’T market themselves to trading customers in the hope that your demand on the system will be at a time that others don’t need it because then the server is probably in the middle of nowhere and more aimed to web hosting and perhaps not that good/responsive with good connections to where the exchange clearing servers are located.

2. A lot of these VPS products have a very basic or customer friendly feel to them so they are easier to sell. The websites and providers that offer them try to make the interface for selling and setting them up as simple as possible, often the process is automated with software, and therefore it is often difficult to find out where your server is ACTUALLY located in the world. Sure, they will say it is in “New York” or “Chicago” or you can choose a supposed location in these cities, and they will say “in the data centre with 2ms average ping to exchange servers” however you don’t REALLY know where they are. They are only giving you a rough idea of where the VPS is located and they do have the right to migrate your VPS around to other VPS servers (without you knowing, it happens seamlessly to your use) wherever they can serve it most economically to make a buck. What if they move it to a server in another datacentre and today you are getting 2ms pings to the interactive brokers servers and tomorrow you get 20ms pings? The server provider wont tell you, you’ll find out yourself the hard way.

I know the physical location of my dedicated server. A dedicated server is a physical item and is not served virtually. It is in a data centre very close to the Equinix data centre in Secaucus New Jersey. No, it is not in the Equinix data centre where the exchange clearing servers are located, however it less than 1km away (getting server space in Equinix is horribly expensive). I am not sure if Interactive Brokers servers are hosted within Equinix NY4 itself or if they are also in a nearby location and have a cross-connect to Equinix, however when I set up my server I communicated with IB to find out their server location and whilst they would not tell me specifically their physical location (fair enough) they did commit to migrating (or having a cross-connect) with the Equinix data centre. At the time of me setting up, they were not yet complete in their migration and they simply said “soon”. Lo and behold, about a month later they did communicate with me again and confirmed the migration/cross-connect to Equinix was complete. My testing at that time to ndc1.ibllc.com (their “primary” server in New York) I was able to connect and have a ping time less than 1ms (so a nanosecond ping time, which is total porn for IT types…). To this day, I still connect to ndc1.ibllc.com as my primary data server in TWS. I am not sure if everyone can get the service or if it can be changed, however in this query of mine to IB when setting up I specifically said that I wanted TWS to connect to ndc1.ibllc.com and not their chicago server cdc1.ibllc.com and they allowed it.

My provider is Interserver and my server is in their TEB4 location. The four core server plus windows 2016 licence, solid state drive, 8GB ram and gigabit network connection costs 152 USD a month (about 220 AUD per month after exchange, credit card fee etc). I cannot find another provider in this area of Secaucus near Equinix with a comparable product, pricing etc. and the service has been flawless. Yes, I had an issue with the initial setup because they installed a hard disk drive when I had paid for a Solid state drive. However I noticed this in the hardware config in windows in the first 24 hours and only when I asked why they said this was a mistake and changed it to the SSD and I then had to reinstall everything again. Still, this was only one hiccup and only cost me 24 hours of time. Don’t mess around with HDD’s on this kind of thing, a SSD is the only way to go to help with system performance.

Whilst this cost may be considered expensive to some, I know others using speedy trading servers and a few other providers that are already topping 100 USD a month for a VPS, so to give me the peace of mind of having my own dedicated physical machine I don’t need to share with others and to have less than 1 ms ping to interactive brokers is well worth the premium in my opinion. I also consider it as this is my investment in my computer. The best of us spend 2k on a computer every year or two these days so again I see this expense as a simple and necessary expense for my trading business. If I’m doing this game full time and can’t handle a 2.5K per year expense I may as well give up now.

Does it actually result in better fill prices and reduce slippage? I cannot really and conclusively say yes or no, however when I was running TWS here on my local machine in Australia I would often check which primary server TWS was using. There was no primary server in Australia for TWS at that time (not sure if they have one now or not, I would assume not). It would sometimes flip between US East (New york) and US Central (Chicago) and HDC1 Asia as the primary server TWS was using. I would ping these servers to check and the ping times were just bloody horrible. I’m talking 220ms (as compared to less than 1ms I now get in NY server). In Australia, yes, kudos to trying to build the NBN etc but we really do still suffer with long ping times to the rest of the world internet backbone simply because of our distance.

I do get the feeling that when I was trying to trade with TWS here in Australia, in combination with this long ping time, I really do feel like the buckets in which you would get fills would be more erratic and in more weird quantities and over a longer period of time. For example a 2000 unit order might have been spread across six buckets and taken two full seconds to fill. Since moving to the server in NY I would get similar 2000 unit orders filled in perhaps two or three buckets (or in some occasions one large bucket) and in the case when split across a few buckets it would happen in hundredths of a second, for example three buckets filled at 9.30.00113 and 9.30.00320 and 9.30.00405. To me, that can only translate to more accurate pricing, lower slippage etc.

p.s. Do not under any circumstance use Commercial Network Services. I tried them in the beginning and they are just bloody horrible. Mysterious resets/restarts in the middle of the session, outages. Speed issues. Poor customer service. Interface to run the thing is crap. Just a bunch of cowboys promising the world and delivering garbage.

[JulianCohen]

Matthew O’Keefe wrote:

Julian do you have a regime/market trend filter? For me, I don’t run any filter on my MR MOC systems. I just run it every single day. Previous building/testing on only being in the market at certain times proved of no real benefit and only succeeded in reducing the number of trades I could get. For me and my system, the more trades and the more you are in the market is how it is showing success.

What else are you doing for deciding your entries? Any other indicators or metrics? How about ranking?

I don’t run an Index Filter on my MOC systems. For ranking I use ROC which I also use on my MR Swing system.

I am in two minds about this having done a bit of testing. One benefit to the RUA universe for me is that it recovers faster and has smaller drawdowns generally…but it’s a bit swings and roundabouts.

[Anonymous]

Do you use anything other than ROC for deciding which positions to take?

Why not be daring and do one day ROC and see what that does to the number of trades hehehehe.

[ScottMcNab]

Matthew O’Keefe wrote:

I see trading a larger universe and having a larger number of positions as a good way to combat this by helping to not have a system that depends on profits from ‘long tail’ trades (especially if these large quantity orders to be filled in a short amount of time at the open are the long tails)..

Can you dumb this down a bit for me Matthew..if the trades are missed because they are off market or occurring due to trades on ECN prior to the primary exchange wont the relative frequency of these events be the same (eg 1%) regardless if doing 10 trades a week or 100? Or is the impact directed solely to having smaller order size so less likely to get a partial fill? Apologies if missing the obvious.

[Anonymous]

Scott, yes, I’m suggesting that having smaller orders is a way to still get fills as they could be filled more rapidly when a price may be available for only a very short time and you need a smaller order quantity to be filled. Whether that is right at the open after the overnight out of market trades have pushed the price around since the previous close, or even within the trading session itself. I am not suggesting smaller lot size would enable you to trade out of hours. I am just trying to suggest ways that help grab any opportunities quickly when the price only pops below your limit for a very short period of time when it is time to trade.

If you have a limit order for 5,000 units versus 500 units and the price drops to your limit price for even just one second during the session then you may just happen to get a complete fill on your 500 quantity. Especially so if you have a nice fast connection to TWS server, fast computer, close enough to the exchange server as possible so it can react as quickly as possible.

I don’t really like it when I see a doji/hammer/spinning top candlestick when performing my post-close analysis and looking at what worked, what didn’t, what fills I got and why, what fills I didn’t get and why, what fills the backtest for the day shows in comparison to the actual trades of the day and why and I think “But that tail was below my limit for the day, why didn’t I get the fill?” Ever since I traded much smaller lots the problem happens much less frequently and when it does I don’t care as much because I am losing a potential gain on a much smaller lot and I won’t live with the fear that “this is the one trade that is going make or break my month”. So it is a practical as well as a psychological helper.

[JulianCohen]

Does having a fast connection to TWS give you a better fill rate? Surely you place the orders the day before, often 15 hours or so before the exchange opens. Don’t they get filled on a first come first served basis or am I being naive? I’m not sure I see the relationship of speed of server to orders that have been placed hours before the open….

I can understand smaller lot size giving you a better fill but not so much the proximity of the server.

[Anonymous]

If placed prior to the open, in TWS, I didn’t think they then got filled on first come first served once the market does open. I thought that regardless of entry time (whether its ten hours before open or 10 minutes before open) the fills occur with those TWS clients that could communicate/confirm/interact with the primary IB server as rapidly as possible, regardless of the order/queue in which they were received pre-market.

Can you (for example) place the orders in web trader, log off web trader, have no interface running at all then log in perhaps half way through the session and see that you had your orders filled? Or does web trader also need to be running/open at the beginning of the session if you want fills?

On TWS, let’s say I’ve entered some orders 10 hours before the session, then you enter some orders for the same symbol 5 hours before the session, upon the market opening you are under the assumption that I should get my fills first. What if my connection to the IB server is super slow and the work/communications between TWS on my PC and the IB servers takes 3 seconds to complete my order. Should your orders then be waiting around in a queue for mine to completely execute and only upon my completion and the resulting market price can your orders then expect to be executed? If this were true, all customers would have to wait for each and every other customer globally for their orders to execute in the correct sequence. What about those customers with super slow connections? How long does the software wait if there is a delay and a customers order needs to be discarded without execution so the queue can advance to the next customer? What if five thousand orders from five thousand customers for the same symbol are in place pre market? They all get executed in a queue and the waiting/completion with each and every customer and their instance of TWS has to take place?

I would assume that such queueing/prioritization would not be possible, and all the aggregate total orders for the symbol that both you and I have entered pre-market, regardless of the time we entered it pre-market, would then be executed as quickly as technically possible in all regards with a combination of both randomness and whichever clients running TWS on their machines can communicate and work with IB clearing servers faster rather than sooner. If your TWS runs a little slower than mine, surely I would get the fills first, no?

I’m not saying this is what happens, I am just talking, thinking it through. In my mind, I couldn’t see why everyone should be queued up and awaiting the execution of each and every users’ individual instance of TWS running around the world. That would just be too slow wouldn’t it?

Like my question on the web trader, if you enter the trades in TWS, close it, then come back and log on again some time after the open do the previously entered trades recover? This is assuming no usage of Smart API with recovery mode by the way. Just pure TWS all on its own. Will the trades restore or will they have been discarded? If they restore, then I can foresee perhaps there is a log/queue such that it can perform your trades for you in your absence offline and regardless of whether TWS is running or not and regardless of how quickly you can communicate to the server. But if the trades don’t restore and TWS has to absolutely be running to enable the trades at all times, then this is another reason I would say it does not make sense that everyone would wait in a queue for everyone else around the world to have their orders filled in sequence.

Again this is just my musings and thoughts on it. It would be great if someone form IB could confirm how it exactly works.

I also used to set up the trades for the next session as soon as I could after today’s close but don’t do this any more. Being in Australia it would be 6am or 7am when the US market would close. This was great because in the morning I could check how it went after waking up, do the analysis and then get ready for the following session, even if it was a good 15 hours away to the next session. BUT, I changed this habit for two reasons…

1. I found that sometimes it would be 5pm NY time and not all the data for all the Russell 3000 constituents would be updated yet. And then sometimes wait another 20 minutes it still wouldn’t be done. Then I’d run the analysis and find in the daily backtest that I should have gotten a fill on something that I didn’t get and I would check other sources of data and find for some reason the daily data from Norgate would show a low price on that symbol which was 30 cents lower than anywhere else I could find. I’d scratch my head and wouldn’t know why. Then I would just forget it and 12 hours later in the Australian evening it would be say 9pm and the US open would be 2 hours away so at that time I would try and update the data again and run the backtest again and lo and behold that symbol that had the weird 30 cent discrepancy is no longer showing as a discrepancy. The low price for the day is now the same as other sources of data and it now aligns with my trade for the day and the backtest is perfect. So who knows how Norgate get their data and clean it, distribute it etc. however I do find that on the Russell 3000 it does take a little while longer after the close for all data to populate and populate correctly (I’m guessing they probably even clean their data, verify against other sources etc and this would all surely take time, maybe even overnight US time). So on this first point I no longer get so desperate to do my data updates and the prep for the next session within only an hour or two of today’s close. I always give it until after the US midnight and usually do it perhaps one or two hours before the US open, usually around 8am US time (11pm Aus time, just before bed etc).

2. I have found that I like to quickly check pre-market pricing for all the entered orders, as close to the opening bell as possible, to ensure none of my limit prices have already been exceed by pre-market trade. I like to do this in the last 10 to 20 minutes before the open, 9.10, 9.20 am etc. Usually by this time all the pre-market orders/trading from others trading the pre-market have already taken place or have already been entered. Sometimes the pre-market price can often be quite different to the closing price from the previous day and in some instances the pre-market price at this time can already be potentially exceeding your limit price on some of the orders you have placed. Depending on the reasoning why the pre-market price is where it is in these cases, it could be completely valid to just cancel those orders because you already know they are not going to recover and make you money within the session if they are already gapping below your ATR based limit price. E.g. previous close was $40, your ATR based exploration and long limit order sees you entering a $38 limit order. Overnight an announcement of some sort took place like a damaging earnings report or some other negative news story and the pre-market price is already sitting at $36. Already you know that upon the opening bell you absolutely are going to get a fill here because of the gap, and it isn’t going to be pretty because if the pre-market price is already gapping down beyond your own ATR based position at $38 then it just isn’t in any way in hell going to recover within that session to the point of making you a profit, its going to gap open lower and just keep sinking. In October, I had six instances where prior to the open there was a gap beyond my limit price so I just cancelled these in TWS and in all six instances this rescued a horrible loss. They were like a -5% gap pre-market which then became a -12% price down for the day. Yes, I know this is wrong. I know this is breaking the rules of my system. However I am still stuck at the moment in working out how to code this intervention into backtesting in Amibroker.

If the pre-market price is gapping down but hasn’t made it past my ATR limit when I am doing this pre-open check then I just leave it alone. My argument is that I’m confident my ATR based calculation on the limit price is a good point at which the trigger should lie. If previous close was $40 and my limit order is at $38 and the pre-market is now at $38.80 then I will just leave it alone and let it ride. Anything below $38 though, I would cancel it.

This is yet another reason why I am so happy I am doing a large number of small positions. Cancelling one small order manually in 120 orders is really nothing in the grand scheme of things. However cancelling one order on a 15 position system could mean losing a chance at a long tail winner so you would probably grimace more at the thought of destroying your system with manual intervention like this.

[ScottMcNab]

Matthew O’Keefe wrote:

Ever since I traded much smaller lots the problem happens much less frequently and when it does I don’t care as much because I am losing a potential gain on a much smaller lot and I won’t live with the fear that “this is the one trade that is going make or break my month”. So it is a practical as well as a psychological helper.

Thanks for taking the time to share this Matthew. Much appreciated

[JulianCohen]

I also wait until midday Australia time to do my analysis. You can see in the Norgate Updater if you click on the Database tab, it shows the state of the update and it is often at interim stage until about 11am our time.

Interesting thoughts on the order queues. I doubt we would ever get a straight answer from IB about howe it works.

[Anonymous]

Yeah I’ve never bothered asking IB because I didn’t think I would get a straight answer. The only solution I have in instances like this is a brute force solution (my dedicated server) to hopefully get my way.

Thanks for the tip on Norgate. Maybe this can help me do the analysis and order placement during the day then just a quick check again prior to the open for my gap-check trick.

[Author]

Posts