This isn't answering the question. You are down 9percent. Is your model, the old one or the new one, supposed to be down 9percent over the same period. Is your -9percent inline with your model or above or below that? You have 5 years of backtested data. So this should be a question you can answer in 10 seconds. If your results are below the model it doesn't invalidate the model if you can discern why. If you are below that it doesn't invalidate the model but could be execution related.
Many of my trades have re-entries, so it's not always a straight bet with my pairs models. I haven't done a backtest on both of them but I do know its win percentage is 88% with a 0% or greater trade being treated as profit even if there is a little commission. As far as the last trade goes, yes, that's how the model ought to perform, but accounting for every single trade there's only one that really threw me out and that was a morning buy where I lost 5.25%, so if that wasn't there, and I was actually following the system through its drawdown based on the pairs system versus exclusive price physics based trades that I wanted in on it's really hard to say where I ought to be. I know I ought to be at least 5.25% higher, just from that, but as far as monitoring both systems the pairs system is a snap. That's a 1 night ordeal, it's watching the price physics system make its trades during the day mostly, so I don't know.
I'm certainly not saying that's the only way, but I don't know of any other method as accepted as pairs trading, even though we get spurts where a completely non-highly-correlated pairs gets featured by a talking head simply because they might be in the same industry. That mistake has absolutely nothing to do with making a pairs trade. The only thing that makes it a pairs trade is the fact that you chose to trade that pair when a real pairs trade is based on known temporal high correlations and very little to do with say a particular industry or sector. It's really a mathematical thing.
I just don't get the attitude, Prado. There's no other way to advertise a newsletter than on a website. Getting subscribers is very important, but I think he does make a lot more than a few grand a year.
So you can't compare your results with the reality. How can you say the win percentage is 88 without a backtest? Without these verifications how do you know the model is any good in reality? One trade doesn't make a system.
I know this is confusing. I have two systems that I use: pairs and physics. The pairs model has a 0% profit or better on 88% of its trades. Physics has a 62-69% win percentage by itself. Combine the two drawdown gets cut in half, and return is actually greater. I know when I did backtest doing that the win percentage was in the high seventies versus 88% of straight pairs trading. The lower win percentage with physics has more to do with the fact that it tries to catch every single possible turn before it happens, and if you're nearly always in the market you have to have a second check by another quantitative analysis to determine whether you should trade or not.
After the last trade I don't think it really matters what I was expecting the first few months. Personally it was going as expected and my model will have a proverbial J-curve so common throughout all of finance. I would say they've been worse than I expected, but that could be because the trend is changing to bull in NQ, but no trend for ES. Oh, MAN! I just looked. We've changed trends from primary Bear to primary Bull. We'll see if we break the high of today, otherwise that's a lower high on its way to a higher low followed by what now will be a higher high.
This is a good time to share my dream last night where I told an investor I had a model that makes 150,000%. It's like your expecting a million from 25. Why not take 25 million and make it a billion?
Okay. Fair enough. It can drawdown 9percent when the trend changes. This type of drawdown is expected. If you are expecting to sell a quantitative model you should be able to explain this stuff. There are a thousand quant models claiming to be the best. Good luck with yours.