Consistently Profitable Trading Strategy! 200,000+ Trade Backtest! Best Results Yet!
HTML-код
- Опубликовано: 14 авг 2024
- While I usually backtest strategies promoted by other videos, for this video I bring you my very own strategy from my own backtesting research. These are the best results yet, and this is an easy strategy to implement, using the MACD, Stochastic, and EMA indicators. Thanks to everyone for the kind comments on my videos!
Music by Coma-Media from Pixabay
pixabay.com/us... Хобби
Excellent job and thank you for sharing. This is very insightful and informative. I appreciate your willingness to allow us to learn from your testing expertise.
Love seeing REAL statistics in data analysis. very very cool
This is the kind of conent I am looking for! Great job.
Ive tested this strategy and it works well on 3 min timeframe with a little twist. You can only buy if 20 ema is above 200 ema.
Winrate 75%
is there a P/L Report for these Trades
I love your videos and your in depth backtesting, keep it up!
I do want to note that I backtested this strategy on the ES and noticed one big flaw. The strategy seems to have worked in the last 10 year bull market but shows consistent losses since the beginning of this year (bear market). I would advice everybody to be careful and do their own testing on stocks, often strategies work differently depending on what kind of market we are in. This might also be the reason why shorts show up negative results in the past, I'm curious if shorts from only the past year would result positive.
I love this comment this is so insightful and I totally agree; i have not been able to find a strategy that is truly not affected by the overall market condition. On the daily tests here I did use 5 year data rather than 10 year for that very purpose, i thought the 10 year data would be so dominated by bull market... but in truth even the 5 year data is mostly bull market, especially since I have to leave a few month buffer at the end to make sure the trade simulations have the time to play out, so it is never testing right up until present day. The other big elephant in the room is that there is no guarantee that past performance predicts future performance at all, no matter how good our results are.
@@seriousbacktester At the end of this video you mention the additional entry rule "price > 200d exp mav" that seems to boost results compared to the opposite. In combination with the comment from King Arthur I tend to conclude the following: While in a Bull market (generally rising 200d mav) it is very profitable to enter "heavily oversold" stocks (below 200d exp mav) as these stocks seem to have a tendency to "snap-back". In a flat or Bear market this technical tension ceases to exist and profitability gets hampered. I "think" (believe, no proof) that technical trading systems need to take care of overall market conditions. My thoughts circle around a trend quality indicator (TQI) applicable for different time frames to decide which regime the market is in.
Why not add a market condition requirement to this test. For example the 10 week EMA must be above the 20 week EMA to go long@@seriousbacktester
what a channel! Keep up the hard work best backtesting channel so far, if I could recommend a few things, could you please show the profit and loss as a graph, with profit being on the Y axis and time being X to see when the strategies were working the best, lastly would love to see you do some crypto backtesting.
Brilliant channel the only one that does comprehensive mass backtesting I'm aware of. I trade the 9ema pullback on the 1 minute timeframe over the major indices successfully but I hide my stoploss behind 50 and 200 EMA, daily pivot, HOD, LOD etc with a R/R of 1 to 1.5. Could you backtest this strategy please but obviously you would have derive the TP/SL from the ATR.
You trade it on stocks ?
Yeah please backtest it
Would really appreciate if you also share the python code... Or at least in your video also share glimpse of code.. So that we may verify/replicate the process
I use some indicators for my trading algoritms: WaveTrend, Chaiking money flow, CCI Stochastic and Volume flow indicator. Very recomended indicators. In case you want the python code I can send you.
please can you send
That's great piece of code you are using... Kindly make one video on how to write bactesting framework
In forex , this will work if all three variable are lined up , e.g. if its above the 200 ema , oversold on the stochastic and under the zero line for MACD . It doesnt seem to work when the currency is trading below the 200 ema
Outstanding!! It would be great to see this in BTC or ETH. Thanks for all you do. Great information
It would be nice to see a crypto strategy :P so i concur
I love your videos! Can you please do some forex and cryptos back test
I assume this isn't leveraged at all. Profits could be much higher if swinging long dated options or swinging futures. Why not do that? Also I'd be interested in seeing a backtest during the recent bear market on ES like another person mentioned.
Wow! Thank you very much. Do I understand you correctly that all the exit criteria need to be satisfied before closing a trade, or do you close if any one of them is satisfied?
Yes, both the macd crossover and the stochastic criterium
Did you add commission in your calculation as well? for the daily candle, 5 years data, 35k trades total or 7000 trades per year or 27 trades per day. if commission is 0.01%, then commission per trade 0.027% per day. net profit per day = 0.045%-0.027%=0.018% per day. So profit per year would be around 5%
for hourly, it would be a loss if you have 0.01% per trade
Thanks for your testing. If we apply this strategy to 1 hour time frame of Forex trading, the return could be doubled because Forex is trading 24 x 5.
How much do you invest in each trade? And could you calculate the kelly criterion to find the optimal trade amount?
You are an Hero
Amazing content backed by real data. Keep it coming. What library you are using for back testing? I would like to learn more about backtesting from you.
Thanks! I use python and pandas/numpy for data manipulation. When i have more time I hope to make a video series on how to code all this but it’ll be a big endeavor, there’s thousands of lines of code I had to write
Wow! This was amazing!
Amazing work thank you for sharing. I’d like to join Peter below and mention equity index futures, specifically the e-mini and micro e-mini CME. Potentially more profit available with a smaller account balance, with win rates of 60+ percent.
Once again thanks for the videos.
Can't wait to see an update on this strategy!!
Thanks for sharing your findings!
Couldnt you just flip all the numbers to enter short positions aswell?
Thank you very much for sharing!!! Appreciate it!!!
A big THANK YOU. Love the content.
Thanks!
Great content as usual
I think the work you put in is great. Unless I missed something it looks like the change in value for the upper and lower band for the Stochastic was overlooked during the setting setup for the indicator. Initially I missed it also.
Love your content ! I was wondering like for entry we need all conditions to be true, is it the same for exit conditions.. all 3 to be true? Or is it any one of those. Would appreciate a revert
Brilliant work - brilliant channel. Thank god for reality!! Question, can you check this strategy…. On the daily chart, long only entry - when all three agree: RSI above 50, Macd signals buy (any level), and Superbuy is a buy. Exit = when RSI falls below 50 and MACD signals sell. The key is waiting until both rsi and Macd agree. I’d be interested in seeing the outcomes.
thank you. Can we get a bot for it? I would like to add this to crypto
Awesome video! Thanks
I love you man! Please make more :P
Hi, if you had to put a 40% stop loss it might point that this strategy is just taking advantage of the bull market we had after the march 2020 problem. Do you have the graph of the profitability of the strategy? Is it a consistent straight line going up or does it show a lot of drawdown? What about 2022/2023 and all the losses in most stocks? If it is a strong winner and the profits graph doesn't swing much this is a serious good strategy.
i think it doesn't swing much, we see on the bell curve result that the random sampling give us an accuracy of more than 95% the drawdown seems to be very low
is there anyway possible that we can get in contact with you? have you considered a discord server? we need some way to message you!
Does this work on forex pairs?
What an amazing video with great content. 🌟🌟🌟🌟🌟Thank you SB. Love your work.
Now, at 6:51 mark, the result shows a 28.973% profit / year. Is that based on the invested $5000 per trade or based on the initial total $10000?
Much appreciated if you can clarify that. Cheers
This is the most interesting strategy I have seen to date. I am a ToS user and tried to duplicate this to test on a couple of stocks I follow to determine if those would also show a profit. What I found very difficult to do is try to alert when all the conditions for both the entry and the exit occurred, but wasn't successful. If you have a list of stocks following them manually just isn't feasible. Do you happen to know of a way to set these conditions as alerts?
Absolutely! Sounds like an idea for a video... i actually do have a thinkorswim video where i show setting up alerts and such for a different strategy, but would be happy to make one for this strategy... i actually currently have thinkorswim watchlists for entry and exit from this strategy on the s&p 500 and s&p 100
Superb strategy... I tested this for Indian blue chip stocks and getting stupendous results... Could you publish number of bars held and avg. amount of capital blocked in positional trade at any point of time
Can you tell me how have you backtested for Indian blue chip stocks?
What max trade length did you considered for the daily and hourly candles simulations, and what datetime period was used in this video?
Awesome video. I am wondering if the volume indicator could be used as a sizing indicator for how big the trade is and whether that changes anything.
E.g. low volume gets assigned 0.5 x typical transaction amount. Normal volume gets assigned 1 x typical transaction, High volume gets assigned 1.5x.
💖Subscribed and liked
In your backtest - did you enter a trade when the previous trade has not finished yet?
First of all: thank you, that you are sharing this professionell testing with us. Question: Did you tried tot test on lower time frames 15min with sell and buy and: buy above the EMA and sell under the EMA?
Love your video!
In the line df = T.macd(df, r1=12, r2=26, smooth=10), what is T? I suppose it is not from finta?
Great question! You are correct, while I use finta for a lot, and it definitely can do MACD calculations, I also have my own class that i've written which contains functions for technical indicators... that's what i imported as T; its pretty straightforward to calculate macd using pandas' built in abilities, and so that function is a carryover from before I discovered finta
What is the advantage of testing on stocks instead of forex? Thanks!
do all conditions have to meet at take pforit? what if the the fast line crosses the slow line but the stoch. is below 65? are ou staying in the position?
Thanks for the question; yes as this has been tested, the position would remain open in that situation
Hey. There is a problem with your backtest however. Stock Market has gone up a bunch in the past bull market and only buying should always give you profits in such a bull market, however your strategy clearly lacks the ability to work in different conditions, such as a bear market... Still really good video and im definetely a sub
Can you share the drawdown?
Great video, congratulations. A question here... whats the c4 condition that show up at 7:06 min? Thks
Great question! This is a filter condition I add to all of the strategies I test; I usually mention it but must have forgotten in this video....
"c4 = (df.volume.rolling(window=10).sum() > 300000/divider) & (df.open > 3)"
condition 4 is basically to filter out low volume stock and penny stocks, it makes sure that the sum of volume over the last 10 daily candles is > 300,000 (proportionally less if testing lower time frames, that's what the divider is), and makes sure the price of the stock is > 3 dollars. Doesn't affect s&p500 index trading much but when I'm testing full market there are a ton of those stocks that are like 30 cents per share or have super low trading volume, and since I actually do trade I want to make sure I'm testing stocks that actually have enough liquidity to trade, and for low price stocks the test can be misleading because you might think you can make 10% profit if the stock price goes from 10 cents to 11 cents, but in the real world if you ever try it, you rarely actually realize the gain. so i stay away from those stocks, and i filter them out of the backtesting. Thanks for helping me clarify for everyone; i'll try to remember to always mention that
How about adding the additional condition that there should be regular bullish divergence in the MACD and/or Stochastic (or RSI) as well as cross over of MACD fast and slow?
Sir, every time I try to back test a strategy it gives an error saying no json token. What is the meaning Sir?
How about some multi-timeframe action? Like trade on 1H, but only if Daily is in uptrend.
If there is a 5050 chance of market going up or down...why doesnt data show shorting is just as profitable, or is this due to the stockmarket on average going up aka long strats will do better eventually?
Hey your videos are awesome !
Do you think filter stocks by financial data (like canslim or petrovsky score.. )this trend following strategy should be better?
I have no idea if we can access those data in the past… perhaps some forwards testing are needed?
1 question i have is what if i trade this on 5 Currency Pares and 3 Indizes. Would i have made 20% on Each Marked ?
first of all thanks for sharing...! good job.... but how about trade time? trading sessions?
HEY MAN , i think u should add price running below a low as a Type of Liquidity grap on the 1h or 4h Timeframe and then use the usual setup for entry , i saw massiv impovement PLS backtest this
Looks like profit/year percent doesn't matter, but number of trades matters. Total profit of 141k trades would be times bigger than 4k trades
Excellent work! I will start paper testing it today and let you now what i find.
How is the paper trading going with this strategy
@@barbarosglobal Failure, same as every one of the 200+ strategies I’ve tested. They all fail given enough trades.
@@barbarosglobal None of these videos will tell you this but every strategy needs to be carefully adjusted for each equity and then adjusted over time. However, when you test them over a longer period they all fail to make any profit. I’ve tested hundreds of strategies and they always fail to make any profit in the long term after you subtract trading fees.
@@SuperheroArmorychannel Hi, sad to hear that this strategy isn't profitable. Have you found a profitable strategy by now?
@@wedeltest Nope, they do not exist. I’ve spent thousands of hours testing and no single strategy works consistently. Perhaps in the future AI could help to create one but all these channels showing profitable strategies are completely bogus.
How would you manage risk here if your loss could be 40%. Would you lose 1% of your portfolio if there's a drop of 40% ?
Great question....the answer is that it all depends how much someone invests, which depends on their risk tolerance... also, remember the stop loss isn't foolproof, an investor could lose more than that.... if someone sets a 40% stop loss but the stock opens the day down 90% due to some horrible news, then that investment would lose 90%
thanks for the hard work. have you tried trailing stoploss? and why dont you backtest forex pairs or cryptocurrencies?
Thanks! Actually I recently tested a trailing stop for this very strategy on some common settings e.g. 5%, 10%, and it just wasn't better than the results here, but will continue to check that out on future strategies. The reason behind no forex or crypto is kinda selfish, its that i don't trade crypto or forex... plus i only have a little forex data and even less crypto historical data even archived
is there a P/L Report for these Trades
could you test this strategy on forex please?
CAn you try this the opposite way to see with shorting would result ind similar results ?
Great setup. Do you use any kind of screener/etc to find trades that'll meet the conditions easier? If so what are the screener filters?
Great question; I personally use td ameritrade's thinkorswim platform which allows me to put these rules in easily as a custom watchlist and then can send emails and push alerts when stocks meet the criteria
Hello, what would be the appropriate thing to do in this strategy if an entry occurs and this stock comes from having touched the stop loss of 40% recently? Better to wait for another opportunity or enter the same?
Hello sir
Can you please backtest this strategy I used 4EMA crossover of 30 EMA in Nifty 50 index SL 10 points
I am an option buyer
I assume you built the simulator yourself? Finding good back testing software seems difficult to find without doing your own coding.
thanks! yep written in python; i have a series where i show how i do it; working on part three now of that series
I came across your video and find it very interesting. Thank you for sharing your results. Are you open to share your code with me so I can make some of the adjustments to it and do further testing?
hola, y en CRYPTO como seria de rentable?
Price below the 200 EMA but only if the daily Timefrime shows an uptrend?
Can I link this with binance so when criteria rich binance automatically place a position for trade ?
Hey man nice video , Can u pls backtest a strategie that includes VWAP and Divergences , this claims to be very profitable
Hi Mr Serious Backster!
there is any chance to do a backtest for this same strategy for selling entries ( opposite rules, and usin 35 level of Stochastic)? I´m so curious to see the results. Thank you in advance!
Best backtesting channel!
I wonder how this strategy would play out for the bull Spx stocks in this bear market?
thank you for the video, does this apply on crypto?
can you elaborate more about entry amount and total capital?is it 100k account that uses 5k per each trade?does it just enters a trade whenever any stock meet the criteria and you still have cash?
Great question... i code it just to take every trade using $5000, regardless of what the running total of gain/loss is, as if you had unlimited money to take $5000 trades, so I take trade 1 for $5000, and if gain $100 then I now the balance is $5100, and then on trade 2 lets say i lose $300, now the balance is $4800... but the next trade will still use $5000.... its not the traditional way to do it, but when dealing with hundreds of thousands of trades, i don't want it to take 200 trades and then stop due to the balance being zero... i want to just let it run to completion so I can calculate the ultimate yearly profit or loss. In truth I wouldn't even need it to use dollars at all, I could just add up percentages of gain/loss for each trade... i just use the dollar/balance more to help me visualize or put in context the results. the more important value is the overall profit/loss percentage per year
@@seriousbacktester another very important metrics are max drawdawn and max loss streak
I can find many strategies that yields for hundreds % per year but they might have a loss streak of 30+ which is aint problem for robot but you as a human most likely would surrender after 10+loss
@@timurzhaizhumov4810 Thats a great point; for these tests even something like drawdown i would need to present as a bell curve of likely outcomes as each person would be executing a subset of the hundreds of thousands of trades... also under the hood my system is not strictly testing chronologically... for example it's not taking March 1 and tallying all the trades on that day across the whole market in order....it's taking the data for stock A, calculating all it's trades, then moving on to stock B and so forth. I am going to start introducing graphs of the running balance, but with the understanding that its not really a chronological balance, its formed by consecutively adding the results of each stock. The noise in that balance line would give an idea of the expected consistency, but won't be exactly the same metric. I hear you on the value of that though, great comment!
I've created an indicator based on your setting but unfortunately it's making a loss. Tested it on the 1hr BTC timeframe for the past 8 months.
great job
Great channel, I have a strategy I’d like you to put to the test how can I send you this please ?
hi SB. can you backtest the supertrend with the dema set at 200... If interested i can send the settings? cheers
thx
What's a stop of 40%? 40% of your all account?
Great content you've been sharing!
I've had most success with quantitative investing, so it would be fun to see if that kind of stockpicking could improve the results: The "tradeable universe" is limited to the top n stocks at certain fundamental multiples. Specifically high p/s outperforms according to studies. High-momentum stocks also tend to outperform the indices: Add 3M+6M+12M price performance. then sort on the compounded value. Higher is better.
Also, some other commenter mentioned that in his own backtests the strategy is profitable only in market uptrends. Would it be possible to include a criteria checking the direction of the index? For example, the index must currently be priced over its MA20 for a trade to be taken that day.
A general request for future working strategies, could you include max drawdowns and sharpe ratios?
Take care
could you explain the stoploss condition in text, you said 40% stoploss, but is that meaning that the snp would have to drop 40% to stop out or the position value has to drop 40%
Great question I should have been more clear…based on entry price…so stop sell if the current price < 0.6*entry price
nice... thank you
Where can you find a backtester like the one your using?
Please clarify when you say 40% stop loss, what is this percentage? 40% of the stock price when taking the trade or 40% below the recent low? Also what is the average risk to reward ratio with this given the profit target is variable? Thanks
I don't think there is an average considering the TP is determined dynamically. The stop loss I couldn't figure out either, so settled on 10*ATR which seems to yield reasonable results across about 20 pairs.
@@Henry-zw2ys Thank you. Would you be kind enough to share your back testing stats such as annual gain and win percentage.
40% of the entry price.
Why don't you do this for forex ? Why don't you like forex ? 11% per year is not that much.
hi, what kind of tools did you use ? and how did you got the historical data for all the stock markets ?
or i guess you coded it yourself ?
what is a 40 percent stop loss?
how about removing EMA completely
Yeah if you are experienced enough then you can
What I'm testing currently is actually removing the entry requirement for macd crossover, just keeping the condition that the macd lines are below zero. On initial testing it gives similar profit but more than doubles the number of trades
@@seriousbacktester this strategy will not be profitable in long run I have tested it before
Sir I have question you tested this strategy on number of stocks... But should not we optimize the strategy just for the stock we are interested in trading...
For instance I like to limit my self to only couple of crypto tokens when trading... As strategies working on one asset does not work on other one...
Or should I backtest my strategy on many assets before applying it onto any one asset?
Or should I keep my self open for taking position in any asset when signal is given.
😅 Too much to ask I guess... But would really appreciate your guidance....
Its a wonderful question! The challenge when limiting to one particular investment product, like a single stock or a single crypto coin would be having enough data to have confidence in the results. I would personally feel more confident if i find that a strategy generalizes well to different investments and different market conditions. Who knows, it may not be possible, and it may be that for all the effort, past performance may not even predict future performance at all. I think that's a question everyone would like an answer to. For example i like the results I've had with this backtest so far, but maybe with the addition of more data it won't hold up; time will tell i suppose. And I hear you on showing more code... i'm trying to strike a balance between people interested in that and people bored by that... i think i'll have to make dedicated videos about the coding aspect but I appreciate the interest!
@@seriousbacktester thanks... And yeah i agree to that. Video format is best and engaging...
For code there should be dedicated videos
@@seriousbacktester I would also be interested in a coding video and I think that many of your other viewers are coders also. Maybe here you could also get into details about the assumptions that you make, e.g. what is the exact criterium for being an S&P500 stock, should it be in the index at the beginning of the test period, the end of the test period or when the trade is entered? This might make a difference to the bullish bias. Fees and slippage are other details that I would be interested in.
drawdown bro ?
how does it apply to Forex?
2.35 million days of trading?
What is the meaning of 40% stop loss?
It means sell stop set at entry price* 0.6. So yes a very very deep stop loss, basically I’ve found this strategy, having an indicator based exit, has a built in loss mitigation mechanism already…you’ll almost always be better off waiting for an exit based on the stoch/macd even if that’s for a loss, rather than letting it hit a fixed stop. So the stop loss is a last resort
@@seriousbacktester ok. Thanks for your explanation and strategy. I’ll try it. GB
The thing that makes no sense to me is you can literally just program a bot to do this and if you truly get these results, its just a money printer right? Am I missing something?
I think the element to remember is there is no guarantee at all that past performance predicts future performance so any money invested is always at risk. That being said, yes if your broker had an API then it’s straightforward to code automated implementation of strategies