Grrr…Idea’s Needed

[Nick Radge]

Terry,
Your concerns might be well qualified, but missing a few fills won’t make a huge dent to the bottom line. Obviously time will tell.

In terms of getting on the wrong side during a takeover, yes, that also a risk, but also one that’s covered in the backtesting and with the stress testing.

Nick

[JulianCohen]

Terry one other thing to consider, well two actually, is that the short system works really really well in a rising market. It’s not there to protect you in a falling market. After all it is a mean reversion system so any stock that is going up “too high” gets sold.

Secondly this is an MOC system so any “pain’ to be felt in holding a losing stock is closed the end of day. One of my short systems has a max trade drawdown of 92% but it is running at 40 positions at 7.5% so the pain is only for a day.

[Nick Radge]

Also consider that many stocks under takeover ‘tend’ to be falling. This system is only shorting rising stocks.

[TerryDunne]

Hi Julian,

I probably didn’t express myself too well, so I’ll have another go…

If you go long on a stock that’s ‘oversold’ however you measure that, and it drops hugely in price intraday, the most you can lose is the amount that you have invested (in your case 7.5% of your portfolio) if the price goes to zero. Upsetting but not the end of the world.

But shorting is different..

A made up example…Stock XYZ has made your top 40 potential shorts. You set your short limit price above the close by whatever your stretch is and that is triggered at the open. So now you are short XYZ 7.5% of the value of your portfolio. In my made up disaster scenario there is a reason why XYZ has moved up as the company announces something incredibly good – lets say a cure for all forms of cancer. The price of XYZ moves up 10 fold (but why couldn’t it be 100 fold) on the news and you have lost 75% of the value of your portfolio.

I know this is extreme but extreme things happen all the time. At least when something extreme happens on your long trade your loss is limited…but not on the short side.

You can probably tell that I’m not likely to be shorting any time soon

[JulianCohen]

I totally understand Terry. Your possible loss is limitless in theory, whereas going long you can only lose your stake.

However by keeping the account size relatively small…for example each position in my Short MOC is worth 0.75% of my equity devoted to MOCs, the potential for ruin can be mitigated. Also running MOC instead of holding overnight also assists.

Good thing about investing is, each to their own

[Anonymous]

I’d like to chime in here because I feel like I’m on the cusp of having something good but I’m not quite ready to run it with live money. Julian, you mention that you are trying to find something that isn’t RSI. Here’s my take.

What Said said in another post about “I am not creative so I always start with a very good ranking first” (or something like that, I didn’t double check it) really stuck with me.
My live MOC system doesn’t work if I invert the rules. I think this is because “stocks take the stairs up and the elevator down” – the route up and down is a bit different and won’t work with my conditions.

So I started to think about ranking. I want to find overextended stocks, but I also want to find extremely volatile stocks, because conceptually I think it is safer to have a short system fill less often than a long system. I don’t want to find one or the other, however, and every attempt I’ve made so far to use volatility inside a condition as a hurdle has not resulted in a good volatility expansion condition.

Enter:
Rank = RSI(5) * (ATR(1)/C)
Multiplying by the daily volatility (found that gem somewhere else deep in the depths of this forum) seems to be worth it’s weight.

For conditions, I only have one… and it’s not even a shoe-in.

With that in mind, to me it seems that due to the system being nearly ONLY ranking, I’m using the R3000 with 40 positions. The better the ranking is, the more I want to have all the best ranked stocks at any time in the system, and not splitting across universes.

Is my logic flawed anywhere? I hope this helps someone!

[Anonymous]

Pre-2005 it is all over the place, but it is not broken. GFC causes a huge drawdown, but other than that, DD sitting at ~10% with the same stretch multiplier.

[JulianCohen]

Interesting.

I have also looked at systems that really rely on ranking purely as the “entry”.

My way of looking at it is that if you have a system that works, and passes all your stress tests, then it doesn’t matter why it works. Use it until it doesn’t (if ever).

The quants at Renaissance are all mathematicians and physicists, not traders. So when they look for a system they are not looking to test an idea based on trading knowledge. They don’t have that bias.

As for RSI, I just wrote a short system that has great metrics and it is based on RSI. However I am using a ranking system that Larry William’s followers would not be using, as they all love Historical Volatility, so I think that will keep me away from competing with them for trades. I added it to my stable.

[JulianCohen]

I just tried your ranking idea Seth…I like it!

[AlexanderLuong]

For ranking, I tend to like:

1. Momentum
– ROC
– (Annualized) Exponential (not Linear) Regression Slope
– RSI
– CCI

1b. Coefficient of Determination = Correlation Coefficient^2
– Momentum * Correlation(Cum(1),C,5)^2

2. Math to combine Momentum & Volatility and make it bigger
– Momentum + Volatility
– Momentum * Volatility

3. Volatility
– ATR
– Bollinger Band Width
– StDev
– Historical Volatility = StDev(log(C/Ref(C,-1)),P1) * sqrt(250)

3b. Inverse Coefficient of Variation = abs(C-MA(C,5))/StDev(C,5) or abs(C-MA(C,5))/ATR(5)
– Volatility + Inverse Coefficient of Variation
– Volatility * Inverse Coefficient of Variation

[Anonymous]

Hey Alex, regarding your point #2 – I have also tried to weigh the RSI and the ATR differently, squaring one or the other to make it “more important”.

Will look at your 1b. idea today…

cheers

[Author]

Posts