I have been working on replicating the result of your NQ futures version but I found my results are quite deviating and I think I have done something wrong. May I ask in your version do you ignore the trade starting at 09:30:00? My trades numbers have gone to around 10085 trades.
Ah I see. Previously my thought was there could be noisy area formed at 09:30 bar’s close and then we could enter a trade there if the signal is valid.
Fantastic article and thanks for sharing. For the calculation of the upper and lower band, did you use any differencing or ratio adjusted NQ and ES futures ?
Thanks! I was thinking the way they made into continuous contract will affect the signal generation process or not if we compare today signals with 1 year later ‘s backtest on today signals.
By the way just wanted to add my encouragement for this strategy - the comment about the slippage wasn't meant to be negative but just a warning for the evaluation. I have been running a tweaked version of this strategy since the paper dropped (also using ES and NQ similar to you) but with 1minute entry evaluations instead of half hour and additional stop and take profit criteria and it has been performing very well even with slippage and costs. Slippage is also a very tricky thing to backtest for as it will vary greatly over time (eg. NQ right now can pretty easily slip 4+ ticks but 4 years ago a slip of 6 ticks or so would be massive and not reasonable)
Great article! During your backtest, did you also only take trades at HH:00 and HH:30 as indicated in the paper? Seemed a bit random when reading it and doing some backtest many times a big chunk of the move is lost due to waiting for those specific times.
Yes, I followed the paper. I also tried exiting immediately (not waiting for the 30’ windows), but that didn’t work. Trading only during these intervals solves the noise problem. I believe using other frequencies (e.g., every 28’) will also work… but need to test
Hi, a bit off-topic, if you don't mind... have you tried testing any SVXY/VIXY strategies, which could be used as a porfolio hedge? I recently came across this article: https://www.diva-portal.org/smash/get/diva2:1447840/FULLTEXT01.pdf I tried replicating some of these strategies with Pine Script (I'm not good with Python yet) but the results were underwhelming. Maybe you could write an article on volatility strategies in the future?
I'm curious what kind of size you are trading and whether you're entering with limit orders? Since you mentioned IBKR and whenever I market into or out of a position with more than a few contracts on NQ the slippage is more like 4-8 ticks instead of 1.
I'm forward-testing it for now. The next step will be to run it in a small account (about $2m). Then, if it delivers, the idea is to scale slowly (I have demand for strategies like these)
I have been working on replicating the result of your NQ futures version but I found my results are quite deviating and I think I have done something wrong. May I ask in your version do you ignore the trade starting at 09:30:00? My trades numbers have gone to around 10085 trades.
Correct. As per my understanding of the paper, it is impossible to have trades at 9:30, because there's no noisy area yet.
Ah I see. Previously my thought was there could be noisy area formed at 09:30 bar’s close and then we could enter a trade there if the signal is valid.
Fantastic article and thanks for sharing. For the calculation of the upper and lower band, did you use any differencing or ratio adjusted NQ and ES futures ?
Thanks! No, I'm just using Databento's continuous contracts (NQ.c.0, ES.c.0)
Thanks! I was thinking the way they made into continuous contract will affect the signal generation process or not if we compare today signals with 1 year later ‘s backtest on today signals.
By the way just wanted to add my encouragement for this strategy - the comment about the slippage wasn't meant to be negative but just a warning for the evaluation. I have been running a tweaked version of this strategy since the paper dropped (also using ES and NQ similar to you) but with 1minute entry evaluations instead of half hour and additional stop and take profit criteria and it has been performing very well even with slippage and costs. Slippage is also a very tricky thing to backtest for as it will vary greatly over time (eg. NQ right now can pretty easily slip 4+ ticks but 4 years ago a slip of 6 ticks or so would be massive and not reasonable)
Great to hear that it's working for you! Question: doesn't trading every 1 minute instead of every 30 minutes lead to overtrading?
Great article! During your backtest, did you also only take trades at HH:00 and HH:30 as indicated in the paper? Seemed a bit random when reading it and doing some backtest many times a big chunk of the move is lost due to waiting for those specific times.
Yes, I followed the paper. I also tried exiting immediately (not waiting for the 30’ windows), but that didn’t work. Trading only during these intervals solves the noise problem. I believe using other frequencies (e.g., every 28’) will also work… but need to test
Hi, a bit off-topic, if you don't mind... have you tried testing any SVXY/VIXY strategies, which could be used as a porfolio hedge? I recently came across this article: https://www.diva-portal.org/smash/get/diva2:1447840/FULLTEXT01.pdf I tried replicating some of these strategies with Pine Script (I'm not good with Python yet) but the results were underwhelming. Maybe you could write an article on volatility strategies in the future?
I'm curious what kind of size you are trading and whether you're entering with limit orders? Since you mentioned IBKR and whenever I market into or out of a position with more than a few contracts on NQ the slippage is more like 4-8 ticks instead of 1.
I'm forward-testing it for now. The next step will be to run it in a small account (about $2m). Then, if it delivers, the idea is to scale slowly (I have demand for strategies like these)
IS THIS AREA BOLLINGER BND ??
No, it is not. The noisy area is thoroughly explained in the paper