I am sure that you have read articles like the one below: https://ca.finance.yahoo.com/news/monkey-beats-stock-market-172803113.html Well, this is a back-test result for a simple trading strategy following two exponential averages crossover. Stops are set on stone before entering the trade. 250 points down, exit. Code: Pending Long Pnl: -758.30 Net Long Pnl: 8803.10 Best Long Trade: 1298.70 Max Long Drawdown: -490.2 --------------------------- Pending Short Pnl: -758.30 Net Short Pnl: -5485.80 Best Short Trade: 4682.70 Max Short Drawdown: -780.14 Market data is in an enunciated clear uptrend, hence the better results for the long branch. Nothing too exciting here, we all have been in front of these results. If the market is in your favor you make money. Big brain algo. Here is when things get interesting, I remove the EMA indicators and pretend that a monkey is trading. Prerequisites for the monkey: Only one position at a time. Enter on a 1 out of 7 chance, meaning that if he tried 7 times to enter a trade, he would be successful once. Why seven? No apparent reason, just like the number. Monkeys seems to like certain numbers. He can distinguish between red and green, green is good and red is bad. Bananas have been used to train colors. Code: Pending Long Pnl: -758.30 Net Long Pnl: 4246.70 Best Long Trade: 1452.70 Max Long Drawdown: -278 --------------------------- Pending Short Pnl: -458.30 Net Short Pnl: 8408.20 Best Short Trade: 4166.2 Max Short Drawdown: -460 For the life of me that I tried to run this monkey algo for about 40 times and can't get a time where the short side is negative. The long side has reduced in Pnl almost every single time but it never loses money! Are we trying to over-engineer our algos when a simple random chance would do much better? Here is a picture of the monkey I had in mind when I was developing this algo:
%% MAKE sure to have sheets of paper notes or notebook ; thus the nickname ''sheet monkey'' could be a key. I like a sunrise /ducklblind ..............................b side; 3 chords + the truth ./Luke Combs/ Loving Song.\ >49,777,777 video views Edit or 49,888,888 video views/ like Rich Dennis noted money is in conservative hands
%% Honest Injun; works well in an up/ Trendin' bull market. I"M walking on Sunshine; + dont it feel good/HEY LIVE In Norway/American Bandstand+ other places >72,777,888 video views/ Katrina [Kimberly]+ the WaVes1985+LOL
How many days / bars were used for your back testing? Maybe the number 7 was just a lucky pre-curve-fit guess? How do other numbers around 7 work? Do 8 and 6 behave similarly? In my experience, the value of the variable you’re tweaking should show a trend in p/l, otherwise it’s likely curve fitting. Random guessing should be breakeven minus commissions. Breakever is better than most traders and is achieved because it lacks the emotional component of marrying a losing trade 50% of monkeys lose 95% aspiring day traders lose. I don’t remember who, but someone on ET once said something like this (and it drastically changed my trading experience) Set you profit target the same distance away as your stop loss AND DONT MOVE YOUR TARGET OR STOP. Long term, you’ll be breakever (minus commissions). If you have the slightest edge, you’re likely to be at least a by little better than breakeven (long term)
I have used about 8 months of data. 7 bars was just the worst case possible, all other combinations produced better results. So I chose that number to get the less fitting one. I was writing the post in a joking mode, so I didn't explain everything in detail. The point was to prove that a random algo can do better than a very simple one with EMAs. When I backtest algos properly I run them in loops with parameters in a range, I choose always the less over-fitting parameters in order to maximize my trust on it. If you look at the profits in 8 months you can easily see that any decent algo would have made a lot more in that period. Your recommendation for stop losses is probably a good solution for a proper algo, but I didn't want to improve it in any way, so I could easily see if a random execution would do better. If I modify the first algo and apply proper risk management with stop loses the monkey should not beat it.
%% YOUR post sounds profitable; uptrending bull market. Even though that can drawdown, that'$ ok + good reason sometimes. SINCE he disclosed ''no...reason for number 7''; its midpoint between 5+8 fibo numbers.LOL Since most of his posts are much more serious[like his nickname ''dd addict''; i did not confuse a monkey with a man LOL. But speaking of addictions ; I'm rightly addicted to 8 or 9oz glasses of water a day.