Plutus is a Script I've been working on for the past several months, and It's finally ready to go live! I would like to have few volunteers to help me forward-test it with different assets, preferably, with any Binance Perpetual contract traded under US$50.
The Script is delivering a Net PnL of circa 350% (0.04% commission deducted from the ~500% gross) in 14 days on the backtest when plotted on the WAVESUSDTPERP 2-min chart. Good results also with RUNEUSDTPERP and OGNUSDTPERP .
To whomever contributes with a constructive (technical) feedback, a better Stop-Loss strategy, other potential (Binance Futures ) pairs where the Script would delivery a 15-Day PnL of 400%+, and MAINLY who collaborate fowardtesting it, I'll grant access to the Strategy (which will have its Privacy Setting changed from Public to Private in 7 days from now) for a period of 2 months.
Please, explored it with different time frames, assets, markets and leave your comments!
Pretty decent script. One issue I can see with it though after observing it over the last 24 hours is that it will print a signal to enter or close a trade, then after I switch time frames then go back to the original timeframe some of the signals its painted completely disappear.
For example I've had it on SOL for the past few hours, printed a short scalpe signal on the 1min, gave me a close signal 14 bars later, then gave another signal to short 15 bars later. I then changed time frames after a few bars had closed and switched to the 4HR and daily TF then switched back to the 1min and the only signal that was left on the chart from the ones I mentioned was the original short scalpe signal.
Has happened a few of times but the signals have usually hit, not too sure how the repainting of signals are affecting win/loss rate but seems minimal.
karegggere
⋅
repaintttttttttttttttttttttttttttttttttttttt
Bagarabumbum
⋅
well after forward testing. i can definitely say it repaints alot. it repaints entries and also exits. dont use it on real trade. i did with small budget. its bad.
adishaz
⋅
did you had comissions and sliipege?
SkyRocketBlue
⋅
@adishaz, Hey! A commission fee of 0.04% (Binance Futures Taker fee) is already deducted from the NET PnL. Slippage is not being taken into consideration given the high return of the strategy, and that all entry and exit orders (including stop-loss) are placed as market orders anyway, so the slippage could be positive or negative, and adding it to the backtesting would manipulate the results once the slippage during forward testing would be regardless unpredictable, but allow a 25% discrepancy of the backtesting results to real-time results. If you still what to calculate with slippage, you can open the "Properties" tab and add it manually, I also have added a backtesting date range input and Stop-Loss input so you can change it in the way that suits you best. I highly suggest forward testing it with real-time bars and not "replay". If you want, feel free to share here what you're thinking about it so far. Thanks!
lasvegasbrad
⋅
Hey, that is NICE. I can sure help you with the grunt work of cleaning up your Settings panel. My God, its making 9K one contract NQ on the 15 second in 4 days, per Pine. I have been playing with REAL strategy orders. Note the fake Pine code ones. The real ones are actually better. I have 2 ways to connect TradingView to my actual, real $$ account at AMP.
CommanderShepard
⋅
@lasvegasbrad, Did you see it repaint? I checked on bar replay and the results become different.
SkyRocketBlue
⋅
@CommanderShepard, I wouldn't advise forward testing it with bar replay mode, but with real-time data. Yes, the code has more than 300 lines, and couple of those lines will cause a repaint, but during the real-time forward testing the strategy will behave exactly as displayed on the backtesting. What is causing the repainting is the function "request.security" and the exits when backtesting. The entry orders will be filled at the bar close only, but the exit orders will be filled anywhere intra-bar, so as the backtest just uses OHLC data, few exits during forward testing will be misplaced if compared with the backtest as in either scenarios those exits would avoid new entries to be opened. Forward test it with real data and you should expect anywhere around 80% of the backtest results.
CommanderShepard
⋅
@SkyRocketBlue, Sorry, but this is a repainting strategy. I let it run in real-time, refreshed the chart and entries appeared on the chart where there was no entry alert at that time.
lasvegasbrad
⋅
@CommanderShepard, oh no, I did not. Good point, MOST of these "great backtest" strats repaint. But SKY has a good answer. I also have noticed this Pine-crap issue in my own testing. You must use barstate-is-real, or alert-once-per-bar stuff.
For example I've had it on SOL for the past few hours, printed a short scalpe signal on the 1min, gave me a close signal 14 bars later, then gave another signal to short 15 bars later. I then changed time frames after a few bars had closed and switched to the 4HR and daily TF then switched back to the 1min and the only signal that was left on the chart from the ones I mentioned was the original short scalpe signal.
Has happened a few of times but the signals have usually hit, not too sure how the repainting of signals are affecting win/loss rate but seems minimal.