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)
A trading associate of mine is creating a system that uses indicators and more importantly price levels which he identifies. The system cannot be backtested due to the subjectivity in assigning the levels. His initial testing has shown many trades that last less than 5 minutes. I told him these must be classified as "scalping trades" and should be discarded from formal performance reports. I said that target price and stop price trump any price levels or indicators that may be used in the trade decision. Am I right ? I guess part of this issue comes down to the definition of scalping.
Can you help answer these questions from other members on NexusFi?
Why must these less than 5 minute trades be discarded from performance report? I don't see what a definition has to do with excluding trades, unless you are saying his tests are somehow invalid.
Yes, I am trying to find a reason to invalidate his forward test results unless he filters trades that are affected more by noise than by any signal, indicator, price level, etc. IMHO a "noisy trade" in ES is one that lasts less than 5 minutes. In other words random price movement is a major factor in the eventual outcome of the trade. To me those trades are not unlike trades entered without any signal or price level with a profit target and stop loss of a few ticks.
Note: I would not discard any trades if his system was programmed and backtestable. However, use of discretion changes the rules for performance evaluation.
If I was your associate and his system looks like it will be profitable from test results then he should forward test it by running it on sim for long enough to get a representative sample of results (as you say it can't be adequately backtested due to elements of subjectivity), ensuring any positive expectancy isn't based on optimistic limit order fills that wouldn't happen live). Then try it live.
Trading is about making money. I don't see how it is really relevant whether you decide to call the trades "scalp trades" or not and whether you think they shouldn't be counted because you personally don't consider that they are being held for enough time.
If he builds a system that has scalp trades he can run that system with other longer term systems as part of a diversified strategy portfolio and ideally smooth, or reduce volatility, in his overall equity growth.
You do not win as a trader, you just get to play again the next day. If that game doesn’t appeal to you then you should not trade. Gary Norden
Yes, that is one fear I have. Since the system will be doing about 20-30 trades per day, the period of forward testing on sim will be about 2 months or 40 days....thus permitting about 800-1200 trades total. That should be enough, right ?
So from your initial post as you say the system can't be back tested, but refer to his "initial testing", I presume he is trading live data on a simulated account. In that case I would use screen recording software such as https://www.flashbackrecorder.com/express/ (the FlashBack Express version is free and works well), have a Depth of Market and a time and sales on screen and record the trading session.
If he has a recording with 20+ of his trades he will easily be able to see whether:
a. The trading simulator is filling the orders instantaneously as if they were first in the queue, or whether it is only filling them more realistically based on how many orders are in the book at that level and how much trades?
b. If he got stopped out was it actually a price sweep where in reality he would have got some slippage but was actually filled immediately in sim?
Ideally he would trade a session on sim, print a copy of all the executed trades for that session then skim through the video and check each fill and tick off all the ones the DOM and T&S show would have worked and remove any trades that look like they wouldn't actually have been filled in real life and see how the two P&L figures compare. Then repeat as necessary. One day's results would be enough to tell him whether it appears to be realistic/on the right track or not.
Obviously I have no idea what markets they are trading, whether using limit orders, market orders, stop market orders but if what I suggested sounds practical I would give it a go and see what the video shows.
Good luck.
edit: Regarding 800-1200 trades or 2 months. As he can't back test it I guess he just forward tests it until he feels he is being consistent and has a clear set of rules about when to enter and exit, and the results are positive. After that I don't see that you gain much for sim trading for a really long time repeating the same thing because the markets change over time anyway so I would then if possible trade a one lot and see how it goes live. Tricky of course if their trading method involves scaling in and out multiple contracts. (and starting with micro contracts if available would fill differently to the more liquid larger contracts, so wouldn't provide transferable knowledge).
Hopefully some of this might be a little relevant or provide some thoughts to consider.
You do not win as a trader, you just get to play again the next day. If that game doesn’t appeal to you then you should not trade. Gary Norden
I did. Totally from experience in the ES and NQ markets. I feel that 5 minutes is the minimum time under normal market conditions to get a change of 5 - 10 ticks which is the target range for expected profit or loss of this strategy. Anything less and one is trading noise.
I can confirm that his set-up with Ninja Trader 8 has been tested with both live trades and sim trades with no difference in the execution of limit or stop orders between live and sim accounts. My set-up is DTN IQFeed with Interactive Brokers accounts and I also see no difference with order execution between live and sim accounts.
Multiple contracts are of course honored, but forward testing will be only with 1 contract. No scaling in or out will be permitted.
Thanks for the feedback, but the real relevant issue is the time-in-trade issue. I am concerned that the strategy's validity will be compromised if there are too many test trades with very short time-frames. The rationale is that trading results for those trades will be based mainly on chance (i.e. noise) rather than any indicator or price level or signal used in the strategy. I call these very short term trades "noisy trades".....the results come from trading the random movements in the market.