auto Backtesting is STUPID... How to actually get a 65% WIN RATE in Trading | Algo Forex Day Trading
This is a Gamestop Stock. How many of you were trading GameStop when it was right here below 10 dollars? Unless you are someone like Roaring Kitty or Michael Burry, chances are you never even looked at GameStop before it made this big move and was all over the news. This is Bitcoin. How many of you were trading bitcoin when it was right here? Probably not many. But when it made this first big move, suddenly everyone and even your dog knew what Bitcoin was. But what do these two charts have in common? Yes, they made a big move.
But after they made a big move, there was a new view about these assets. Few fundamental traders who traded GameStop when it was below 10 dollars had a view that GameStop Stock was undervalued, it was heavily shorted and the price can make a big move up. When it did make a big move in the upward direction and was all over the news, most people who never looked at GameStop before also had a view that price can make a big move in the upward direction because it is heavily shorted, so let me make some money in the upward move.
And after this big move when Bitcoin was discounted, chances are many of you bought Bitcoin with a view that it can make a big move in the future. Now watch this. This is a normal trading chart. It doesn't have crazy moves like GME or Bitcoin, but it is something you will see almost every day. In these kinds of markets, most of us look for trends. If the price is already moving in one direction why trade against it. In
an up-trending market, we don't buy at the top, we wait for the price to give a discount, or in other words, we wait for the price to give a pullback. On the Trading Rush channel, in the strategy tested 100 times series, I tested many trading strategies 100 times, and most of them were trend following strategies. The MACD strategy that got the highest win rate of approximately 62 percent with a 1.5 to 1 reward risk ratio, was also a trend following strategy. When price made moves in an upward direction, we use the MACD indicator to find the end of the pullback. Remember what I just said, we use the MACD indicator to find the
end of the pullback in a trend. Since the MACD strategy got a 62 percent win rate, sometimes, I will receive an email from one of the viewers asking something like, Hi there, I saw someone use an auto back tester with the MACD strategy and ATR stop loss, and it didn't have a good win rate, any thoughts? Or something like, I coded the MACD strategy with 2 times ATR stop loss, but after auto backtesting 1000 times it is only getting around 38 percent win rate, what am I doing wrong or different? I usually reply with. Actually, instead of saying what I reply, I will show you something even better. Instead of relying on some random guy who used an auto backtester to test a strategy 1000 Times, but doesn't know the basics of trading or even understands the strategy properly, In this video, let me, who actually uses and made money with the MACD strategy in the live market, test the MACD strategy more than 1000 times, to prove you why auto backtesters are a hot pile of garbage at the moment unless you are actually looking for a strategy that works with trading bots. This is the win rate MACD strategy got by manually backtesting in the year 2019-20.
This is the win rate the MACD strategy got in the year 2020-21 when we tested it again in the latest market structure in the Awesome Oscillator video. To keep things simple, let's just say the win rate of MACD strategy after 100 trades is 60 percent. Both times, the strategy was tested on the same pair but on different market structures and still got a 60% approximate win rate. I will talk about why I test all the strategies on the EUR JPY pair in a moment. But now watch this. Let's test the strategy again, but this time, let's switch the pair. Let's test it on USD CAD. First, we will test it 100 times manually,
then we will test it again but with an auto backtester. Let's see what happens. Now some beginner traders think that backtesting manually makes the signals biased because you can see the future price. That's not true with the indicator strategies. The strategies I have tested 100 times on the trading rush channel are indicator strategies. Most indicators calculate their values using the closing price of the candle, and once the candle is formed, even your cat knows that the indicator will look exactly the same in the live market and in the past data. The only time you will take biased signals is when you backtest a strategy that doesn't give an entry signal on its own. For example, the V WAP Trading Strategy. In the V WAP strategy, you wait for some kind of reversal pattern near the V WAP line. So
while backtesting it in the past data, you will take a lot of biased signals. And that's why I have not tested it 100 times yet. The same thing applies to any moving average strategy where you buy at the bounce or wait for some kind of pattern to form near the moving average. But to avoid any confusion, I modified the Beep Boop indicator that I created in the AI trades video to give MACD entry signals. This is what it looks like. When it is green,
it is a long MACD signal, and when it is red, it is a short MACD signal. I'm looking at this modified Beep Boop indicator to avoid taking any biased signals even though there are none. And after testing the MACD strategy 100 times manually on the USD CAD pair, it got a 58 percent approximate win rate with a 1.5 to 1 reward risk ratio. There is not much difference between the 58 and 60 percent win rate, but watch what happens when we test the strategy again on the same pair, same market structure, but with an auto backtester. Here's how some beginner traders backtest the MACD strategy with an auto Backtester. They will tell the software to buy and sell at the MACD crossover with the zero line rule and will set the stop loss by multiplying the ATR value by two times. Now at first glance,
this setup looks all good, but there is a big problem that is overlooked by beginner traders. If I run the auto backtester, you will see that it actually got a lower win rate even though the market structure was exactly the same. It got a 47 percent win rate which is 10 percent lower than what the MACD got after manually backtesting it. In the previous videos, I explained why the end of a pullback in a trend is really important. And when I tested the MACD strategy 100 times on the Trading Rush Channel, I tested it exactly how I trade it in the live markets. If you are a subscriber of the Trading Rush Channel
you have probably seen some of the live trading videos, and if you are supporting the channel on Patreon, you have seen many different MACD setups in the Live Trade Analysis Posts. In an uptrend, since the price has a higher probability of moving in an upward direction, we look for any buying opportunities we can get. But one of the best places to buy in an uptrend is near the end of the pullback. But no one knows where the pullback is going to end.
But if you know there is a good uptrend going on, you can use the MACD indicator to find the decreasing downward momentum and increasing upward momentum because when the pullback ends, the downward momentum decreases, and upward momentum increases. That's all the MACD indicator is doing. It is not a holy grail and is not rocket science. But when I take a long entry signal with the MACD setup, I like to set the stop loss below the pullback, because I'm trading the end of the pullback and not the MACD indicator. When you use the ATR value to set the stop loss, it is no longer the same MACD setup that I trade in the live market or tested 100 Times on the Trading Rush Channel. If you are a long-time subscriber, you probably remember that at the end of every video I used to say that, if you use a different setup or change the strategy a little bit, you will get a different win rate. I said that in every video because that was important, but then YouTube said no, you can't say that in every video. Well, when you can't say these important
things every time, beginner traders make mistakes like changing the most important part about the strategy and then say that they are not getting a good win rate. Remember, just because one uses the same indicator while backtesting like someone else, doesn't mean they both are using the same strategy. Some even go far as making a video or posts about it and misleading everyone because of their ignorance. Remember that around 90 percent lose money in trading, so don't trust every random guy you see on the internet. 90 percent of people on the internet have no clue what they are talking about. This also applies to me. Now I have shared live trading videos and have even shared live trade analysis on Patreon that shows I do trade what I talk about on the Trading Rush Channel. But don't blindly
believe everything I say, listen with an open mind, and then use your own head to see what makes sense and what does not. By making this a habit, you will not only learn more about the topic you are interested in but will also avoid the 90 percent who don't make money in the live markets. Not everything backtested with the auto backtester is bad. But you saw how an auto backtester can give such a bad win rate, but that's not the only issue with an auto backtester. Actually,
there are many issues with these kinds of auto backtesting software, and even though I am a developer and have created multiple trading indicators, I mostly avoid using it for one big reason. if you remember, at the start of this video, I asked you how many of you traded or thought about trading GameStop when it was right here. When one of the viewers emailed and asks me to give my thoughts or opinion on a video where some guy tests a strategy 1000 times with an auto backtester, I usually try to explain the market with a coin example. The market is not like a coin toss. When you toss a coin, the probability of a fair coin landing on heads is 50 percent. If you flip it 100 times or 10 thousand times, if you test it 10 years into the past or 10 years into the future, the coin or in other words, the subject will always remain the same and the probability of the coin landing on heads will remain 50 percent. But the financial market is not a
coin toss. Markets are always changing. When the stock market was crashing last year, most people were making money by selling the stocks. In this crash, most of your long strategies or trades would have lost. But wait a minute, the Forex market doesn't make a big move like
this. No, but even your dog knows that market stays in a range most of the time. Sometimes the range market lasts for a long time. So what happens is, when you test a strategy 1000 or more times with an auto backtester, it not only gives a bad win rate as we saw in the previous test, it gives a horrible win rate because. Actually, instead of saying why. I will show you why. These are the win rates I told you to remember before. The MACD strategy got a 62 percent win rate in the year 2019 on EUR JPY with all the MACD rules, check out the MACD videos to learn more about the MACD strategy, then got a 60 percent win rate without some rules in the year 2020 on EUR JPY, and now, in this video when we tested it 100 Times on the USD CAD pair, it got a 58 percent approximate win rate. But watch what happens when we manually
test it 100 Times on the GBP JPY pair. As you can see, it got a 44 percent win rate which is way lower than the MACD strategy got on EUR JPY and USD CAD. I will explain why in a minute. But watch this, if I use an auto backtester on the same market structure, it gets a 36 percent win rate which is still lower than what we got by manually backtesting. So question number 1. Why did the MACD strategy perform so badly on GBP JPY but performed
well on USD CAD and EURO JPY? The answer is simple. If I bring these two charts up, what do you make out of it? If you say the first one is trending and the second one is not trending, then you are right. The first chart is USD CAD and the second one is the GBP JPY. The MACD strategy is a trend-following strategy, and if it is a trend-following strategy, even your cat can tell that trend trading strategies don't work in the range markets. But if we
test a trend following strategy in a range market and if the strategy gets a bad win rate, is it the trend following strategy's fault, or are we just ignorant fools? The GBP JPY was moving in a good trend at one point, but then most of the time it was moving like this, and this, and this. There is a big difference between these two charts. Here on the GBP JPY pair, price is making a big move in one direction, then staying in the sideways for a while, then a sudden small move in one direction, and then staying in a range for the most part. Now tell me, if your strategy is designed to work on this market structure, how can you expect it to make money where the market is not even moving for the most part. In this ranging market, the manual backtesting win rate was low, but the auto back tester win rate was even worse. This not only shows that auto backtesting consistently gives a lower win rate than manual backtesting with a strategy like MACD strategy where the stop loss is supposed to be set using the pullback, but it also shows how ranging markets can affect the win rate of a trend following strategy. Not all Forex or Stock charts are the same. The biggest mistake beginner traders make
while backtesting a strategy, is that they test the strategy in exactly the same way on every single pair or stock. If you do that you will get different results from pair to pair and stock to stock, because some pairs move better than the others, and some stay in a range more than others. But that's not all. If you test a strategy 1000 or more times on the same pair like USD CAD that is currently in a good trend, you are basically saying that the pair will always stay in a good trend and will be or was never in a bad range in the last 10 years. But even your neighbors' dog knows that that's not true. If a pair or stock is making a good move right now, doesn't mean it made good moves every year for the last 10 years. Some
traders will auto backtest strategies on 10 years of data and will say that it doesn't work. How can you expect the Forex or Stock market to remain or move the same way for 10 years? In the last 10 years technology has improved so much. Nowadays Algo trading is becoming a common thing, more people have access to a computer and a faster internet, and thanks to discount brokers a lot of traders can trade Forex and Stock Markets with just a few clicks. If I can make an AI or my computer to trade on its own as a joke, imagine what
can or probably is being built by someone with big pockets. So if someone tests a strategy 1000 or more times on really old data, it is a good idea to run in the opposite direction and use your own head to look into the future, because the future of trading is filled with Algo's and AI. Just yesterday, I saw an Ad on Reddit that said something like, get a stock recommendation by our powerful AI. For some reason they had the comment section enabled, and the top comment was Recommends Garbage stocks. So in other words, the AI is not that
smart yet at recommending stocks or probably is, but they are obviously not going to share the cake with everyone. Question 3. How to backtest properly and how to actually get a high and realistic win rate? You see in the live market, no one is going to use a trend following strategy in a range market. But we won't know it is a range until it stops making new higher or lower lows.
Before a range market starts, one or two trades will be lost. So when I test strategies that I actually trade in the live markets, I like to backtest it exactly how I'm going to trade it in the live markets to get a realistic win rate. Because remember, when you trade in the live market, you have a view on the market. You believe the price is going to go up, go down, or is in a range. But while backtesting, you don't backtest with a view,
and since most of the time price is in a range, you end up taking more trades when the market is moving in the sideways direction, which is not something a trader does in the live market. So to solve this problem, I like to test strategies on charts where the price is moving like this, and not like this. I test strategies where there is a little bit of range so we lose few trades in it, just like we would in the live markets, and most of the price structure should be trending, because I'm testing a trend following strategy. If you blindly take
1000 or 10 thousand trades with an auto backtester, and test trend-following strategies on charts like these where the price stays in a range for the most part, and if you don't have a view on the market, that's not trading, that sounds like gambling with few extra steps. In the how I got an 85 percent win rate and an entire green month video, one of the first things I said that got me no loss in an entire month, was the view on the market. Before using indicators to make things easier, I was trading with price action only. And after trading for thousands of hours in the liive market with no indicators, and more thousands of hours with indicators, you develop a skill to spot strong levels on a chart, and your market view gets more accurate. And that combined with a good win rate strategy like MACD, you
get a win rate as high as 85 percent and an entire green month. 85 percent win rate is really high with a good reward risk ratio, and it will eventually fall to something around 60 percent, but right now the win streak is still going strong. Maybe I will make a video explaining my trades and how I found them, subscribe to the Trading Rush Channel and ring that notification bell to see that.
Question 4. How to take higher probability trades in the live market and how to identify and filter charts that are not worth trading? Actually, I have already made two detailed videos on this topic. Check them out to learn more. Because if you can't identify the range, you will end up taking more trades in the range and blame the strategy. Here's what happens when someone who does not even understand a strategy, removes one of the main components of a profitable strategy and then backtests a trend following strategy in the range markets 1000 or more times like a gambler with no view. This is not something you should do.
Since MACD does not spam you with signals in a short period of time, and since TradingView has a limit on how far you can go back in the past, the auto backtester is finding around 130 trades on each pair. But as you saw at the start of this video, auto backtesting gives a lower win rate when you don't use a pullback stoploss. I use the pullback stoploss in the live markets, and I have been using the pullback stop loss since the very first video on this channel, and even explained why the end of the pullback is one of the most important points on a chart, and the auto backtester proves it even further. Almost all pairs are giving a horrible win rate with the auto backtester.
Understand and Use ATR like a trader, not like a gambler. I have already made a video on how to properly use the ATR indicator, check it out to learn more. And don't believe every random person wearing a suit on the internet, because most people don't make money in the live markets. That's all. Like the video if you liked it, Subscribe and ring that notification bell, to see proper win rates of strategies tested 100 Times on the Trading Rush Channel. Support Trading Rush on Patreon and get trade opportunity
signals and see In-depth Trade Analysis. Thanks a lot for watching.
2021-03-29 20:02
