Question: How best to ensure that new strategy being implemented is performing as expected. Background: I have implemented a a new trend following strategy that trades a portfolio of assets. So I set up an excel to track profits and loss, various stats, as well as using my iPhone for daily monitoring. Since this is a new strategy, I did the following. First, I back tested the results for 20 years. Then, paper traded the strategy for several weeks. Now, I have been trading this new Portfolio strategy for a short time. The back tested strategy software gives buy/sell signals with an equity curve and standard metrics. But that is based on past results. In real time the goal is that the strategy being tested is profitable and giving the expected results. What I have been doing besides looking at basic profit and loss, etc., is running the back testing software and monitoring the equity curve to monitor drawdowns, % wins, % losses, etc. to see if my new strategy is behaving as tested. Does this sound right? Thanks, Larry
what platform? most back testing software is a scam and makes assumptions to render the best results in a back test. what parameters did you pick in the optimization run? if you did like most you picked the most successful settings. you should pick settings that are not quite optimal looking for smoothness of the equity curve rather than the max profit settings. when you pick the most profitable settings it will most likely always fail in real trading. what is the profit factor anything less than 1.25 or greater than 3 will harbor mistakes. how many trades total in 20 years drastically skews results and makes a system either better or worst as a rule more trades more dependable fewer trades less dependable. what markets are you trading, this can also drastically skew backtesting from real world. nuff for now, truly accurate trading models are a daunting task made to look easy by omission of facts and details that the software will assume on behalf of it's user.
I usually get more nbrs (on all of them I typically get average and median) draw down% draw down velocity (did the backtest reach the max dd % after 3 month but the real time test hit it in two weeks) % winning days in a row (usually up to 5) Then I often normalize the data with VIX and SPY - so did my drawdown happen when the vix was 20 but the backtest one (same percent) the vix was 80 If i'm currently in a draw down and it hurts I try to check how often over a X year period I typically have these types of drawdowns I also don't look just at the 20 year backtest - I try to get the 20 year test and also break it into 5 year periods - there's usually a lot of variability in the 5 year comparisons
Two parts: Audit the details. This lowest level of examining every trade to make sure it is doing what you expect. (The tendency is to think that a system's high level outputs will give an accurate assessment.) Then I would do as DeadDog says, highest level. There is a difference between validity and soundness. Every professional tester knows this. If you need to look it up. Don't confuse the two! The lowest levels are for validity of the system. The highest level (P/L) is for soundness. FWIW, I was a professional tester with a formal education in verification theory.
It sounds like you have taken several steps to ensure that your new trend following strategy is performing as expected. Backtesting the strategy and paper trading it can give you a good indication of how the strategy might perform in live trading, and tracking various metrics such as profit and loss, drawdowns, and win/loss ratios can help you monitor the strategy's performance. One thing to keep in mind is that backtesting a strategy using historical data does not guarantee that the strategy will perform well in live trading. Markets can change, and the strategy may not adapt well to those changes. It is important to continue monitoring the strategy's performance and making adjustments as needed. There are also other ways you can monitor the performance of your strategy. For example, you could compare the performance of your strategy to benchmarks or other market indicators, or you could use risk management techniques such as position sizing and stop-loss orders to manage the risk of your portfolio. Overall, the key is to be proactive in monitoring the performance of your strategy and making adjustments as needed. This can help you identify any issues or areas for improvement, and ensure that your strategy is performing as expected. *chatGPT3
Long winded response... lol Traders, You gave me a lot to think about... Here is some feedback. You can kill me if you like? LOL Mark Brown Questions: "What platform? most back testing software is a scam and makes assumptions to render the best results in a back test." I am using WealthLab. I like it because it can help build strategies with a Wizard for basic technical indicators and I can add custom programming for more critieria (hard to do... lol) "What parameters did you pick in the optimization run? I picked parms considering: - I want profitability, but I balance with higher % wins, so that I don't get too discouraged during draw downs. Also, I chose based on 20 years. I think this helps a lot, because I think my own analysis skills are improved by looking back over a longer range. Also, equity cure in a 45 degree angle is what I am shooting for, but now, also looking at higher yearly APR%. "what is the profit factor anything less than 1.25 or greater than 3 will harbor mistakes." My profit factor is much higher than 3, but I thought that was because I use such a long trading period with longer holding positions as my goal for trend following. I would think this is good? I don't understand why this would be bad. "how many trades total in 20 years drastically skews results and makes a system either better or worst as a rule more trades more dependable fewer trades less dependable." 180 trades with % profitable of 65%, I focused on this, so I try to get more wins. Also, I personally want to hold trades a long time. I want this to be a long term trending system. "what markets are you trading, this can also drastically skew back testing from real world." I am trying for diversity, but this is tougher. My universe is Indexes with 20 year history that behave towards my testing (curve fitting... for sure), but I am hoping 20 year history better protects me. Deaddog wrote: "Look at your account balance", "Are you following your strategy exactly?" Yes to account balance. I am entering on a limit order from night before, so this is different than buying the market open. Bad_Badness wrote: "Audit the details. This lowest level of examining every trade to make sure it is doing what you expect." I have created an Excel. I need improvement on this. This is still in Flux. Did not use this before. This is a learning experience. "There is a difference between validity and soundness. Every professional tester knows this. If you need to look it up. Don't confuse the two! The lowest levels are for validity of the system. The highest level (P/L) is for soundness." So true... I wish I learned this when I went into Swing trading years ago. Could have saved me $$$. Age/Experience and losing $$$, has its benefits sometimes... lol. 2rosy wrote: "There are also other ways you can monitor the performance of your strategy. For example, you could compare the performance of your strategy to benchmarks or other market indicators, or you could use risk management techniques such as position sizing and stop-loss orders to manage the risk of your portfolio." I am using my Benchmark as you have stated. I have used Position sizing ( using sizing based on % of equity), with a Portfolio to lower this risk. This is not where I would like it, but I get the profitability to be where I want it. I really like those double digit profits (APR) that people tout. I have never got that. However, I have lowered my expectations to "beat my market benchmark and use compounding of Profit % to help over the long run. Also, with new strategy, trying to limit "wiping out, so I can trade again". Wish I could put a stop loss or trailing stop in. I have not got enough profit with these in back testing. You guys really make me think. Much appreciated for your time. Thank you, Larry