You might wonder why I decided to trade this particular system. What research did I do to show it would be good?
Basically, I came up with an idea, and just ran it through the data. No optimization (although I did look at some different stop levels and profit targets, my initial "hunch" for values were near optimum, so I changed nothing).
This wasn't your typical run of the mill Tradestation optimization. It took me a solid 2 weeks of testing, both in Tradestation and in Excel.
The graph below says it all. It shows the walkforward and real time (both hypothetical) results for this system, assuming 1 contract per trade.
What I like about this system/ why I decided to trade it:
1) Nearly 7 years of historical trade results, all walkforward tested (not your traditional backtest). If you don;t know why I like walkforward testing about 1,000 times as much as backtesting, just go to my website and look at my articles.
2) Approximately $200 average profit per trade, even after commissions and slippage (Side Note: I assume $50 round trip commissions and slippage. In reality thru April, it is more like $29. So, that "juices" performance by 10% or so. I get this through working and timing the orders a bit).
3) Steadily increasing equity curve. It looks good, but is not without its faults. It looks "real" to me, which is what I look for.
I'm curious what others think out there. If these were your research results, would you trade this system?
Hi, Kevin! It's Beau.
ReplyDeleteI wouldn't trade it because it's not robust enough for me. The models I have hit $120k after two years, in their walk forward, optimizations, and backtests. The drawdown is also half of what you're showing in your walk forward.
Appears robust, though, but I'd find something that could make way more than $120k over 7 year periods on 1 contract.
Also, tradestation data is known to be inherently flawed, so using that crap it's amazing you even get this result out of it.
Can you explain why Tradestation data is "inherently flawed?" There are issues, to be sure, but I've never heard anyone say "inherently flawed" before.
ReplyDeleteI see their data spew crap for years. They're volume bars are no good, because if they were, my e-signal database I use to optimize would work with the exact same scripts and parameters if tradestation's data was any good.
ReplyDeleteI remember the minute based charts watching from Fidelity and IB that they are not outputting the same data for whatever reason, and that makes a big difference for some of the strategies. TS handling of ticks especially is questionable because they don't filter bad ticks like they're supposed and claim they do. At least those were the problems I have with them. I haven't seen where their data ever matched anyone elses, and it is probably a problem on their end.
I've compared TS data with CSI data, and the 2 match up pretty well. But, I stay at the 1 minute level and higher (because TS only provides 6 month of tick data, which means you can't really backtest tick data).
ReplyDeleteI've never used Volume bars, so I can't comment on that.
I personally found many issues with E-Signal data.
Sometimes I'll be looking at real time data from 5 different sources (TS platform, TS Futures 4.0 platform, E-Signal, PFG Best and Advantage Futures), and they may ALL be different!
Regardless of data source, though, one thing is certain, IMO. No historical database is perfect.
Just catching up with your blog... That equity curve looks great to me so I would trade that system (provided equity is enough to handle drawdowns of the magnitude shown x 2)!
ReplyDeleteThanks MBA. I appreciate your comment!
ReplyDelete