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)
What’s the right process or best approach to backtestng trade setups and strategies
I’m new to trading futures and would like to develop my skills by replay. Is there a systematic approach to practicing or back testing setups and trade strategies? I’ll take the time to practice but was wondering if there’s a structured process that best simulates live trading and more importantly enhances the learning process. T
Can you help answer these questions from other members on NexusFi?
As rleplae says, but if interested say in discretionary day trading, then in replay, after initial practice and trying out of ideas, try to be realistic and use a comparable account size, daily loss limit and number of contracts that you intend to or are likely to be able to use when you go live.
Most replay accounts are unlimited in these respects so if you practice without loss limits say and trading many more contracts than you would be able to you are wasting your time in unrealistic and ultimately unhelpful practice. It is very tempting in replay for example to take big losses and double down on trades to dig yourself out of drawdowns in the desire to be green at the end of the day and justify that by saying the extra hours spent replaying are beneficial due to the added screen time trying out ideas. In reality the lack of discipline or testing of a specific defined methodology is quite often just teaching bad habits and impulsive trading.
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
@kevinhpchan Okay replay is not backtesting. Backtesting is where you have fully defined your rules and tested them historically. As for the structured process, you will have to create that. The best simulation of live trading is trading in the simulator on live basis. There are small inaccuracies in most replay engines, unfortunately. In addition, even if the replay engine were accurate, you still have the problem that it is difficult to incorporate non price information or price information on different frequencies. Notice that most traders have adopted the "statistical" perspective on markets even though statistical properties of markets are not very reliable. Many of the large traders, instead, have adopted opportunistic or speculative mindset. As an example, the early adopters of Bitcoin weren't thinking statistically, they were buying opportunistically with an incredible risk to reward ratio. I think that both thinking styles can compliment one another but in the replay you can only practice on statistical basis (unless you have an engine that can also replay the news for a day-- would be nice).
It is what you create. However, if you want to practice then I recommend reducing the number of variants (things that can vary). Most day traders focus on the first and last hours of the session. Given that, focusing on those areas where you plan to trade will be more productive. By reducing the variants, you can improve your confidence of being able to translate the performance into a live account. For example, trading with a max 3 point stop loss is going to be easier to translate into live trading then setting your stop based on your discretion on the fly. Make as much of your trading as invariant as possible is the basic goal. If you can trade well under these invariant conditions, you use realistic to slightly pessimistic fill engine, and go live you should see similar results provided the market doesn't change. As for enhancing the learning process, careful documentation and serious review of your trades will help with that.
One other aspect that needs to be developed is market cognition. The mind of a trader should be active, accurate, and clear during trading.
Thank you so much for the useful advice & recommendations. I’m currently learning how to read Charts using the Whyckoff methodology so I want to learn to apply that in daytrading. After which, I will add in indicators and new setups I’m learning (e.g MBoxWave).
The reason why I’m interested in backtesting with historical data is that I can compress more practice time. I will simultaneously also do live simulation.
A couple questions. I’m still evaluating between NinjaTrader & Sierra Charts. Any perspective how each would fare in replay?
I can't speak too much to Sierra, but I can give you some info about NinjaTrader.
The replay system uses level 2 volumes and level 1 volumes to help determine when and if you should get filled. So it is not a simple HLOC price only back-tester.
But there are a few things to know:
1. The Market Replay assumes 0 latency for executions. So when you submit your order, you are immediately in the queue with no lag between: Submitted > Accepted > Working. This chain can take anywhere from 100 to 500 MS for most retail traders depending.
2. The Live SIM uses around 150 MS latency for executions. So it is in the ball park for some retail traders.
When auto-traders try to reconcile their performance between the two, this is the largest factor that makes Live SIM results so bad compared to Market Replay (0 Latency = perfect pricing / timing / reacting, etc.)
In terms of process, what others have said is really what you should aim for. Try to solve for one variable at a time, and try to cut down on doing too much that can blur your understanding of cause and effect. At the end of the day what you are looking for is a bet.... And this bet needs to be able to be tested in isolation and then measured to see if it has merit. Once you muddy the waters by testing too much together, you can't really draw any legit conclusion about what your edge was. I'll give you 1 easy example. Let's say that you think you have a directional edge. When the level 2 data has a specific setup (One side is stacked heavier than the other), you think "a ha, I will go short, or long, or whatever" So you test this 1 variable. But then you also use a larger stop loss and a smaller profit target with every trade. After 100 trades, you find that your system seems to be working. So here is the question: Was your edge, the directional (level 2 stuff) or was it the risk / reward ratio? You will never know... Because you had too many things going at once. So you will need to deconstruct this into multiple simulations, and then test each possible edge in isolation to first see their impact independently. Eventually you can combined edges, but at first when you are exploring if there is an edge or not, it is best to try 1 thing at a time and see..... This all backs to middle school science lab stuff. Same concept, and it works.
No free lunches unfortunately.
Best of luck
Ian
In the analytical world there is no such thing as art, there is only the science you know and the science you don't know. Characterizing the science you don't know as "art" is a fools game.
Not for Algo-traders. It runs the same code in 1x as it does in 1000x speed. There is no performance / accuracy difference.
For manual traders.... It's a bit like playing Tetris I suppose. I mean you can play level 30 on the original tetris where the blocks fall so fast that the game is over before you click the first button but why would you....
If you are trying to do this manual then maybe stick with 1x speed during the us cash open / close when the market moves crazy fast, and when the market slows down, maybe move up as high 25x speed..... so simulate the cash session's fastest points, but I see no reason why you would ever play it faster if your a manual trader.
Ian
In the analytical world there is no such thing as art, there is only the science you know and the science you don't know. Characterizing the science you don't know as "art" is a fools game.
@kevinhpchan The benefit to speeding up the market replay is that you can see how longer sequences unfold. So you can see how general structure unfolds. There is a benefit. On longer time scales, speeding up time can also make it easier to trade because one is able to "prime" better for the unfolding patterns versus in real-time the sequence may unfold so slowly so as to go unnoticed. The cost is that you lose the ability to experience how that process unfolds in the real-time, the self-doubt, waiting, etc. Also, if your personality is such that you like to place more bets then speeding up time can paradoxically improve performance because you can make more good bets within a certain psychological block of time. Example, if you are trading on a 1 minute chart, over a single trading hour, if for every 1 minute you could take a trade and you only wanted to trade the best 20% of the those 1 minute blocks then you would have 12 potential bars to act on within 1 hour. If you were trading 1 hour charts and you wanted a correspondent number of opportunities at the same selectivity level then you'd need to look at 60 hours of market activity. If the opportunities were evenly spaced, you would need to wait an average of 5 hours between trades versus 5 minutes.
Generally speaking, at this level of detail you are going to have to do your own work to determine what makes sense for your trading. I might add, very likely, the easiest way to outperform 70% of futures traders is simply to reduce your leverage.
