50 day back test....
OEC/Tradestation EasyLanguage Strategy
- Thread starter masterm1ne
- Start date
You have to simulate all limit orders with "trade through limit" price movements. This means that if you place a limit to "sell next bar at 1125.00 limit", the price high of the next bar has to be 1125.25 or greater for you to actually get filled. You will find that this will make a substantial difference in your backtest, i.e. you'll be filled on all of the losers and only on some of the winners.
The best way you avoid this problem entirely is use only next bar at market orders with 6.25 per side slippage for the ES.
Nice work in theory is as far as you can go right now.
You need IMO a MUCH MUCH MUCH larger sample size. And laying 10 lots in the ES is too much risk for a $10,000 start account size. Welcome to the next 100,000 hours of the biggest headache of your life.
You need IMO a MUCH MUCH MUCH larger sample size. And laying 10 lots in the ES is too much risk for a $10,000 start account size. Welcome to the next 100,000 hours of the biggest headache of your life.
From what I can see in the orders tab on my back test, some orders are canceled because the limit price is not touched. I don't use mkt orders cause the backtest always shows entry at the low/high at the entry bar, instead of a predetermined price, like in a limit order.
I agree, I am going to open an account with TS as soon as possible so that I can backtest on more data. In theory, wagering 10 contracts on a ES trade is hefty, but what if you win 8 times out of 10? I am currently risking about 5% of my trading account. What's the part about 100,000 hour headache about? Lol yes I have spent a long time developing and revising this, but then again "in life you get a result equivalent to the effort you put forward." I will probably spend much more time on similar stuff as well. It is just money after all.
I am betting I can repeat similar odds as the results show over the last 90 days. Remember past performance is never an indicator of future success therefore the past 1,000 years of back test data technically can't help you anymore than 1 day can.
Let me blunt here; you're in for a rude and painful suprise.
There are people here with a tremendous amount of experience (myself included) telling you that what you are doing is flawed. Statistically speaking, in terms of a low sample size - you need many more trades in different regimes to justify the effectiveness of any model. The regime can change tommorrow and you will be screwed, but, if your model is robust, i.e. it was developed using a very large sample size, you'll make it through the drawdown and maybe even slightly positive after enough samples.
Execution-wise, in terms of using limits, and, now by admitting that you don't understand that you need a trade-through to get a fill on a limit order, i.e. a limit not touched and canceled is not the same as a trade filled at the limit price when the high/low of the bar was the limit.
Until you acknowledge and understand the two issues above, stop developing and spend a lot more time doing decent research. Or, you can remain stubborn about what you already have and give your money away. Since you're choosing the latter, please do something useful with your money rather than throw it away; take the amount you're willing to lose and donate it here:
http://www.kiva.org/
Look at the trade executions on the backtest. Almost every trade will have a drawdown, so you know your entry limit orders get filled. For every trade that doesn't have a drawdown, my experience with backtesting is there will be very few, you know you may or may not get filled, so just delete that profit from the overall profit of the backtest to get an idea of its affect.
Maybe testing it in something like Ninja which has more realistic fills would be a good next step instead of risking real money.
Maybe testing it in something like Ninja which has more realistic fills would be a good next step instead of risking real money.
Interesting that OEC margins seems to be very reasonable.
You might try channeling your displaced anger energy on something more productive.
Calling me stubborn is pretty pointless. I have already admitted that I know there are flaws in the back tests and I know that in order to get fills you have to have the price go through the limit.
Using your advice with market orders changed the profit factor by 0.1, not to mention the problem with the orders going in at exact h/l of the bar. Clearly if I notice the strategy isn't profitable, I would abort it.
If anyone has an account with tradestation and has the ability to test this strategy on multiple years of data and post the results I would appreciate it. Obviously you get my code in return.
Im a Tradestation user and can perform the backtest and post the results. send me a pm so I cant run the test.
regarding OEC, how do yo get to the ez language interface that enables one to type in ez language strategies which OEC runs? All I see is the Auto X add in which allows one to route order files from tradestation software to OEC via email. With this, it seems that one still needs a copy of tradestation.