Thank you both for a great interview. Would Probability Cones be your preferred method or tool for deciding when to shut off your systems? Have you considered testing variations on placement of the probability cones or potential adjustments with re-optimization and overall effectiveness?
I'm conflicted about Monte Carlo. On the one hand, the idea that the ordering of results can give one insight into the range of outcomes and drawdown makes sense. On the other hand, monte-carlo seems to be just a function of win-rate and avg trade result, so how much additional insight is monte carlo really providing beyond those 2 metrics? In general, I would think high win-rate systems would fair better in monte carlo relative to low win rate systems, but I'm not sure that is representative of how those systems would relatively perform in the future.
@@AlgoTradingWithKevinDavey Thanks Kevin. As an alternative or supplementation to monte carlo, do you have an opinion on using a noise generator to build different input time series and then generating different results by running the system on each of those different input time series? It seems like one could also use these results to characterize possible drawdowns and build probability curves. I haven't explored how this would be redundant or complimentary to monte carlo.
![005 Kevin Davey discusses system development, the best systems, backtesting and more [AUDIO ONLY]](http://i.ytimg.com/vi/TpmDzRFIu0k/mqdefault.jpg)
![005 Kevin Davey discusses system development, the best systems, backtesting and more [AUDIO ONLY]](/img/tr.png)
![012 Ernest Chan on quantitative trading, drawdown, automated trading and more [AUDIO ONLY]](/img/n.gif)
Kevin drops alot of great trading insights in this one, paste your #1 takeaway below.
Definitely seems better to be more conservative using a Monte Carlo then to avoid it because of not accounting for volatility clustering
awesome video, this guy rocks !!!
thanks, glad you enjoyed it
Thank you both for a great interview.
Would Probability Cones be your preferred method or tool for deciding when to shut off your systems?
Have you considered testing variations on placement of the probability cones or potential adjustments with re-optimization and overall effectiveness?
glad you enjoyed it @jason
Thank you Andrew and Kevin!!!
glad you enjoyed AW!
These videos are done so well keep it up!
Glad you like them!
Thanks, Andrew! You really drive home the import points with your guest!
thanks Hugh!
A true eye opener.
Thanks, glad you found it informative
I'm conflicted about Monte Carlo. On the one hand, the idea that the ordering of results can give one insight into the range of outcomes and drawdown makes sense. On the other hand, monte-carlo seems to be just a function of win-rate and avg trade result, so how much additional insight is monte carlo really providing beyond those 2 metrics? In general, I would think high win-rate systems would fair better in monte carlo relative to low win rate systems, but I'm not sure that is representative of how those systems would relatively perform in the future.
I find it very useful for drawdowns, determining how big they could be and how long they could be. That makes it worthwhile for me.
@@AlgoTradingWithKevinDavey Thanks Kevin. As an alternative or supplementation to monte carlo, do you have an opinion on using a noise generator to build different input time series and then generating different results by running the system on each of those different input time series? It seems like one could also use these results to characterize possible drawdowns and build probability curves. I haven't explored how this would be redundant or complimentary to monte carlo.
@@davem8285 - I know others do this, I have personally never had success with it.
Dumb Question.
Instead of a running a monte carlo sim, couldn't you take the back tested results and run that into a streak calculator?
thank you!
You're welcome Panda!