I Gave AI ₹10 Lakh to Trade For 30 Days

I gave AI my real money, 1 million rupees, and for 30 days it chose every single trade, and this is the final PnL . 1 million. You can get a decent second-hand car for that amount. I gave that to a chatbot. By the way, I haven't told anyone at home yet. So please, don't forward this video to my mom. But listen to the scariest part. Last year, a company gave six AI models $ 100 each. They gave them real money to trade on crypto. Claude lost 31%of it and ChatGPT lost 63%. ChatGPT is great at writing, but in trading... let's just spare its reputation here. So why did I still give Claude 1 million? Because I didn't give it to him to trade the way they did. I gave it five rules, and these are the five rules you should see. Before you trust AI with your money. So you will see its best trade, its worst week, and the day my own rules stopped it from trading, then the final numbers, and at the end, I'll tell you if I'll give it money to use again. What's up guys, I am Siddharth. You are watching Trading with Siddharth. I have been trading for 10 years and I teach you this every week on this channel. So, I wasn't just watching the AI. I was checking every single trade of it. Like a teacher checking notebooks. My red pen was ready. Let's start with the rules. Only one market will be used, that is Bitcoin futures. In simple terms, whether the price of Bitcoin goes up or down, you can make money both ways. Without buying the coin, only one chart will be used, the daily chart. Claude only sees one new candle a day and it can take a maximum of one new trade per day. One trade a day; before every trade, it had to tell me three things. Where it will enter, where it will exit if it’s wrong, and where it will take profit. In trading language: entry, stop loss, and target. And then come the risk rules. Sort of like a mom's list for kids going to a hostel. Maximum loss per trade is 5,000 rupees, which is just half a percent of 1 million. And the stop loss will be slightly beyond yesterday's high or low . And after taking the trade, it will never move. "Let me give it a little more space.""Let me move the stop loss. " That is not allowed. And more money will never be added to a trade that is in a loss. And the trade size will never be larger than the account. Meaning we will not take borrowed money . This is called leverage. And this is what blows up accounts. And in case of three consecutive losses, we won’t trade for the next three days. A direct time-out will occur. Why so strict? Because last year researchers conducted a lab test. When they gave the AI trader wrong and misleading information , its trades jumped from 47 to 391. Without rules, AI simply doesn't stop. That is why I decided the breaks myself . Look, the entire prompt with all five rules is free. I will tell you where to find it at the end. Now, let’s start with the first week. And what it did after the first three trades surprised me even more than the trades themselves . So on day one, Claude did nothing. Just read the rules. It’s the first time I’ve seen someone read the terms and conditions. So day two, first trade short, short at $ 78,500, stop at 79,600, and the next day target hit at 76,800. On day four, took a long, long at 77,600, our target hit at 81,800, and on day five, shorted at 81,000 and target hit again. Three trades, won all three. So in 5 days, we were 28,780 up. Now, if you have traded, you know what happens next, right? We won three trades in a row. Now we start choosing the color of our Lamborghini and trade immediately on whatever candle we see next. But Claude did the opposite. It didn't take any trades for the next three days and that day it had a note. No clear setup after three wins. So we wait. Wait after three wins. I am going to put this line as my phone wallpaper. But this patience was about to be tested because the next three trades went into a loss. On day nine, a long was taken at 79,000, stop loss hit. The next day, shorted at 78,500, but Bitcoin went up. Again, stop loss hit, and day 11, long at 78,300, stop loss hit once again. Three losses happened, so now for the next three days. So, the account fell from 1,28,000 to 1,12,000. Look, do you know the pattern of a common trader? First loss, it happens. Second loss, there is something wrong with the market, and third loss, it's all an operator's game. Now look at the size of each loss, about 5000, every time a 5000 stop, the rest is broker's fee. Now this is what rules do, my friend. How much the loss will be was decided before the trade. Look, most people don't make big losses from one bad trade. The biggest losses come from the trades taken right after. In anger, in revenge. I’ll take this trade back right now. This is called a revenge trade. And on Day 11, that trade was right there on the chart. After the third loss, the price jumped a bit on the chart. My heart said that now it’s definitely going up. While looking at the chart. Long at 76,900 and stop at 76,600. But the rule said that after three consecutive losses, no one will trade for three days. So Claude didn't take that trade, and I checked later what would have happened. If it had taken the trade, the price fell to 76,530 that day, so that trade would have been a loss too, and a back-to-back fourth loss. So for the next three days, that is the 11th, 12th , and 13th of September, nothing was done, zero trades. The AI just sat there, like that uncle at a wedding waiting for the buffet to open. And honestly, I didn't like it. I was at a 16,000 drawdown and my own rules stopped me. It’s like my own house is locked and I can’t even go inside. Now some of you might be thinking, Siddhant, 30 days is too little. This is luck-based, right? You are right. 30 days is very little. No one wins a World Cup by winning three matches. But look at this data. SEBI, the referee of the stock market in India. They conducted a study on intraday traders. Those who had made over 500 such trades in a year. 80%of them lost money. Claude took 14 trades in 30 days and sat idle for 16 days. So the question wasn't whether AI is smart or not. The question is, does it follow rules better than you? Because what happened after those three empty days is the reason for this video. So Day 15, timeout is over. We are back to the same risk. Need to recover a 5,000 loss . We didn't use a big size for this one . Right? So we took another long trade. Target hit once again, and then this happened. The best trade of the entire test. We took a long at 75,500, with a stop at 74,900, only $ 600 away, and a target of $ 81,000; the target didn't hit that day or the next, but Claude held on. If it were me, I probably would have exited at a $ 200 profit and gone out for panipuri, but on the third day, Bitcoin surged to $ 81,000 and the target hit. Risked 5,000 but we earned 40,900, More than eight times the risk. Then three more: a long at 81,400, a short at 86,000, a long at 83,300, and we made a profit on all three. Six wins in a row. 121,971. This is the total monthly profit, which came right after the timeout I was frustrated with, and honestly, the last two trades ended in losses. So, the loss was of the same size, around 5,700, not even 1 rupee more than that. And friend, this is the real game. Our average win was about 16,750, and the average loss was 5,634. The win was three times larger than the loss. This isn't about brains, bro. This isn't about intelligence. This is discipline. So, what did we make in the end? Here is our final P&L. Plus 122,583. From 10 lakhs to 11,22,583 in 30 days—that's 12.3%, and this is after deducting fees. The fees were around 8,362, which is 6%of the profit. Now, in a SEBI study, even the winning traders were paying 19%of their profits in fees and charges. Meaning, if you trade too much, it's a festival for the broker every month. Look, the AI is the same, but the rules are different. So, the result came out completely opposite . But there is a twist here. If you had bought Bitcoin on day one and forgotten about it, doing nothing, you would be up 7.4%. This is called "buy and hold." So, the AI beat the "buy and sleep" strategy too. And the maximum loss in any single trade was only 5,834; remember, this is before taxes—in India, crypto gains are taxed at a flat 30%, and you can't even deduct losses from profits. So, should you let an AI trade for you? Not alone. The AI didn't win because it's smart. It won because it simply cannot break the rules. It has no ego. And it doesn't have a friend sending "surefire" tips on WhatsApp. And after this video, I will use one thing every day. Before every trade, I paste my plan into Claude and ask one question. "Which rule of mine is this trade breaking, bro?" And it stops us before we take those angry trades. So, no more revenge trading. So , will I use it again? Yes, but only with my trading journal. So, the exact prompt and all five rules I gave to Claude are free. The link is in the description. Now, don't copy the trades . Copy the rules. And if three losing weeks have scared you, then you definitely need to watch this next video. I have a complete risk system so a bad week never wipes out your entire account.