Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
I finally have a strategy that backtests very nicely. Ironically, over the last month or so I have programmed many strategies that I thought would work well, which kept failing miserably. I’ve spent many hours and late nights hacking away in NinjaTrader over the last few weeks coding different strategies to no avail. So yesterday, just out of desperate frustration I pulled a “George Costanza”
So when my algorithm says to buy, I make it sell instead of buy, and when it’s in the conditions I thought would be good to sell, I buy. And wouldn't you know it I started seeing profits. I added a few tweaks and played with different settings and now I have these results below.
I don't like losing that much money on just one trade, but the retracements can take a while and they need room to wiggle to achieve high success ratios.
This is on the June ES contract, Mar 1 – Jun 1
Works well on forex contracts too, here’s 10 contracts on the Sept 6E
That’s 87 winning trades and only 1 losing trade! I actually had it making more money in some cases, but at the cost of a lower winning percentage ratio (85-88%). I'd rather have a higher probability of success than more money with higher account volatility.
I’m looking for huge moves in price and, instead of chasing it like I normally would, quickly go in the opposite direction if the move is statistically irrational . It all lines up nicely with some of the things I picked up recently in “Fooled by Randomness”.
I’m going to keep backtesting farther back in time and with other instruments. If that goes well then later this month I’m going to start forward testing this and, if the testing goes well, plan to go live with this on small positions in August in full auto-trade mode (I don't even know how to do that in NT yet)
Sorry for not giving more details, you don't have to believe me if you think I'm full of it, making it up, or just trying to create hype so I can promote something (which I'm not), I just had to share my excitement with someone. I'll post updates as I proceed.
Exactly. Because experience tells me you've simply run into an NT order bug with the backtester and in reality your strategy is a bomb. Sorry to be the bearer of bad news. Hopefully I am wrong, but pretty sure I'm not.
I am taking a guess you are using limit orders and are either getting filed on the same bar, so the OHLC order is unknown, or it could be the SetProfitTarget bug, or any number of other bugs.
Use market replay and let us know. Be sure to test multiple days.
If you set it based on CalculationMode.Ticks and the names of the entries are not unique then you can run into problems. Either name each entry with a unique name (like "long " + CurrentBar) or use CalculationMode.Price.
The more dangerous "issue" (not a bug, but a design limitation) is the entries and exits on the same bar, NT doesn't know the OHLC order and it doesn't assume the worst.
I have also found that using "Strategy Analyzer" for back testing and optimizer produces results that do not bear out in live or replay. I basically use back testing and optimizer to hone in on a strategy and then use replay to actually get the correct results. One other "KEY" is the data feed. As an example, I was running a strategy I am working on, on two PCs concurrently live today. Same strategy, on both. One was getting the feed from IB and the second from Zen-Fire. The IB feed triggered an ill timed trade and ended the day with a loss. Before starting, I even made sure that the IB pc was synced to atomic clock and the T&S sales were within a second of the Zen-Fire feed. (I guess IB sends out the tick data without a time stamp so the PC clock is used, and Zen-Fire includes the time stamp). Anyway, running the same strategy on two different feeds ended up with a huge difference in results. For back-testing purposes (replay), I think it is critical to have time-stamped data so realistic results can be found.