Said Bitar Daily Journal

[SaidBitar]

Julian Cohen wrote:

Said Bitar wrote:

here is an example attached

yellow columns are the optimized ones
green row is the one that i picked

I see what you mean!
If you run an optimization on just one criteria, say Index Filter only, is it a better spread across the results?

the thing is that i am using the index filter as switch between two values
so the thing that needs to be tested are the stretches but the values of the stretches by themselves doesn’t mean much without the switch

maybe i need to fix the stretch and not optimize it and only change the index filter with time

i will check

[SaidBitar]

So i checked the ASX MOC system to see what results it is giving and compare it to backtested results
Live trading results:
Start date 3/13/2017
total trades 161
Win% 46%
Total gains 0.3% (the account value changed in the middle of the duration but based on the starting value it is 1.2%)
in $ values
Gross profit 2648 AUD
Total Comm 2098 AUD
Net P/L 586 AUD

Backtested results over the same period (Note that in this period there was not a single day where the trades where more than the allowed number so in theory the results should be similar)
# trades 193 (there is 32 Trades difference not all but most of them were the low of the day)
Winning % 50.3%
%returns 3.21%
Total Commissions 2316 AUD
Net profit 1604$

[Nick Radge]

So if some of the missed trades weren’t low of the day, what were they?

[SaidBitar]

I believe twice it was my mistake i imported the orders to the API but did not send them to TWS this sums up to 6 trades

[SaidBitar]

I started Tracking Fills on ASX systems
8/29/2017
ASX MOC
GEM.au (Missed Trade) Entry on the low of the day, damage is 1.86% on the trade value

[SaidBitar]

I was thinking about MC simulations tonight and i am thinking to do it in a bit different way it sounds logical to me but maybe it is not.
so it will be done in two stages the first one will take all the possible trades and record them as a separate ticker so i know on each day how many trades were taken
second stage is to run the normal backtest multiple times same as we do the MCs now but it will read the stored number that represents the number of trades on the specific day and if the number is less than or equal the maximum allowed trades then it will not skip any trades but if the number of trades is higher than the allowed trades it will start skipping the trades based on the difference between the total number of trades and the maximum number of the allowed trades

for example
Max allowed trades is 20
Day 1 the number of all the possible trades is 15 then the skipping percentage is 0
Day 2 the number of all the possible trades is 5 then the skipping percentage is 0
Day 3 the number of all the possible trades is 45 then the skipping percentage is 56% (45-20)/45

i know that it will not guarantee 20 fills in the backtest but i think the results will be closer to real life combinations

[RobGiles]

Said,
Is the gut feeling here that a MRMOC system is not as suited to the ASX as, lets say, the Russell1000. Would a MR system that held positions for more than 1 bar work better, or is the low of the day issue going to persist regardless? Or have I completely missed the point here?

[JulianCohen]

Rob Giles wrote:

Said,
Is the gut feeling here that a MRMOC system is not as suited to the ASX as, lets say, the Russell1000. Would a MR system that held positions for more than 1 bar work better, or is the low of the day issue going to persist regardless? Or have I completely missed the point here?

No Rob you are on the point. We are concerned that due to missing trades at the low of the day on the ASX, something that seems to occur with regularity, as opposed to the US markets where it doesn’t seem to occur, that we are not getting real life results that equal the backtest.

This also applies to a standard MR system as well. I have a number of missed trades on my ASX MR system which will also affect the real life results.

The only way to decide what to do is to keep empirical data. I haven’t been doing that, but am now recording all my slippage on both US and AUS markets. Once I have a decent amount of data I will decide what to do. I’ll probably decide at Xmas as that will give me 3 months of data.

One concern on the ASX is that because the brokerage is higher than the US, by trading it and getting a lower than expected result, we are enriching the broker as much as ourselves. If that is the case it might be better to put those funds into a system that gives better results

[RobGiles]

Thanks for that Julian. Why is the ASX a market that you guys have been targeting to trade a MR system? Is it so that you can keep capital working around the clock? If so, would any of the Asian markets be any good?

BTW, is there anywhere in the forum where MR & MRMOC systems actual annual performance figures have been recorded for the US market?

[SaidBitar]

the reason that i target the ASX is that i know i have clean data for it, if premium data will have other markets i would look at them but they only have these two. And honestly i got spoiled with their nice clean and well organised data

regarding the missed trade it was the low of the day seems that it happens often so i am checking them for that day i had 3 or 4 other trades that were OK all of them so it did not make big impact but sometimes i believe it will

[SaidBitar]

Said Bitar wrote:

I started Tracking Fills on ASX systems
8/29/2017
ASX MOC
GEM.au (Missed Trade) Entry on the low of the day, damage is 1.86% on the trade value

8/30/2017
No missed trades
score 1(missed)/7(total)

[SaidBitar]

AUG 2017
(Wow Aug is over man days run very fast)

Account 1
US MRV = -5.48%
SP500 MOC = -1.28%
ASX MOC = 0.85%
Acct 1 Total = -4.23%

Account 2
ASX MRV1 = -0.07%
ASX MRV2 = 0.83%
SP1500 MOC = -6.03%
R1000 MOC = 0.16%
Acct 2 Total = -4.23%

Account 3
US WTT (Closed Positions) = 2.11%
US WTT (Open & Closed Post.) = -0.95%

Account 4
US MOMO (Closed Positions) = -2.06%
US MOMO (Open & CLosed Post.) = -2.13%

All Accounts = -2.83%

Some modifications I did to the systems that i am trading and to their number:
I will reduce the number of systems and i will keep only one of each type:
US
1* Trend Following
1* MOMO
1* MRV
1* MOC

ASX
1* MOC
1* MRV

so the total number of systems that are removed is three

another thing during the last month i found out that the divergence between the backtest results and the live trading results came from the fact that the Selection bias was out of the limits and this happened because i was measuring it based on number of days i.e. 90% of the days i am capturing all the trades but when i calculated it based on number of trades the disaster appeared so the numbers were ranging from 65% to 75%

anyhow now everything should be better and things should be aligned with the backtest results so hopefully profits will start coming.

[SaidBitar]

Last month i found some interesting reports that are provided by IB

these reports can show the performance of the accounts and compare them t SPX and other indices also it shows details about the account and its performance

these reports can be accessed from “account management beta” in the snapshot

[SaidBitar]

SEPT 2017

Account 1
US MRV = 0.98%
ASX MOC = 0.87%
Acct 1 Total = 1.52%

Account 2
ASX MRV = 1.66%
US MOC = 0.44%
Acct 2 Total = 0.95%

Account 3
US WTT (Closed Positions) = 0.02%
US WTT (Open & Closed Post.) = 3.05%

Account 4
US MOMO (Open & CLosed Post.) = 6.39%

All Accounts = 2.52%

[JulianCohen]

Horrah! Positive territory

[Author]

Posts