Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
Hello FIOs!
Once again I'm hoping some of the experienced traders can share some advice.
When trying to gauge the expectancy of a manually backtested system, is it enough to penalise the results of every trade by 2 ticks i.e. assuming limit orders on entry and exit wouldn't have been triggered and manual market orders had to be used?
Assuming small size (1-5 contracts) on ES during the European and US session (~0700GMT - 2100GMT) and at least trying to stay away from big news + not holding overnight, do I have to additionally include slippage?
Thanks for any advice & have a great day!
Tom
Can you help answer these questions from other members on NexusFi?
1 tick slippage per side is reasonable. BUT, the bigger issue is in the accuracy of your manual backtest. When I used to manually backtest, I had a tendency to cheat with the results. Sometimes it would be very subtle, but whether it is skipping trades, accidentally missing trades, or assuming fills that got touched but not filled, or something else, I personally never found manual backtest to be accurate.
It is kind of like how when you look at an indicator on a chart, usually it looks like it is great, and it is picking great trades for you. But on close inspection, the losers (which you barely saw at first glance) appear and usually take over...
As @kevinkdog mentioned, you need to think about how you are going to ensure accuracy of your manual backtests. I have never really had issues with this, but I would just start at a certain point and work my way through the data, only observing the results at the end. Same thing that Kevin mentioned with the indicators is very applicable here - our mind will tend to focus on the instances where they work and ignore instances where they don't. Automated backtests avoid the issue completely.
With regards to slippage - sure 1 tick per side sounds reasonable, but if you just use limit orders you do not need to use any slippage on your winning trades. However, that does not mean that all trades were executed, so a better method to test this would be to insist that the market passes 1 tick beyond your limit order before it executes at your price. For instance, if you want to buy CL at 52.95, you should only consider that the position executed if price drops to 52.94. You would use the same logic with your profit targets.
Regarding stops, start by using 1 tick slippage, then increase it to 2, then 3 to get an indication of how sensitive the system is to slippage. You can't control slippage, but knowing how sensitive a system is to slippage gives an indication of robustness.
You then also need to consider the impact of news events. I am now referring to the EIA reports for CL, but when those were issued, slippage of 50 (or more) ticks can happen. Whenever you backtest over such an event, it would be prudent to assume the absolute worst result, i.e. if the market moves 50 ticks in an instant, odds are you will be stopped out at the bottom therefore your backtest should make this assumption.
I agree 100% and while I have been trying to be very strict when going through the charts, I will check the initial performance from that against a bar by bar replay using historic data and trading it manually "on the right edge".
This will then hopefully expose any bias I may have introduced during the testing on the charts.
I failed to note that on when recoding the backtesting which is why I now asked about the 2 tick penalty i.e. simulating market orders vs. limit orders.
Still not the same, as you mentioned so for further tests I will have to include limit order vs market order result to see which performs better overall.
But the final plan will be anyway to execute on Orderflow data (Bookmap probably) which (in theory) should only improve the results of the purely chart based backtest...
As I mentioned, I'm currently looking at ES and will try to avoid news, so 50 tick slippage is hopefully out of the question