Parallel Testing Crypto Trading Bot Strategies
Exploring Strategy Optimization
If you want to know if your trading bot configuration is creating MORE profit or LESS profit, it is time to run parallel bots testing different settings and test over the course of a year.
That is what is in store for me this year.
I have been running crypto trading bots for about 3 years, and so far have only found the kind of returns I wanted in 2023 when testing Martingale bots on KuCoin and Pionex.
Before that, I was testing mostly Grid bots with little satisfaction.
I had different Grid bots going for almost 2 years. On CryptoHopper, KuCoin, and Pionex.
I haven’t been parallel testing the Martingales because I have been excited enough with my returns.
But I did parallel test the Grid bots the entire time I was running them.
When I run parallel tests, I try (even though not totally possible) to keep it as strict as a scientific experiment when controlling for variables.
- Always test the same trading pair (unless the entire test is comparing different trading pairs on the same bot configuration)
- Invest the same amount
- Keep configurations as consistent as possible. (if cross testing using different platforms and tools, sometimes the settings just aren’t quite the same, so you have to go with what you have)
At the end of 2 years with grid bots —My results were :
About 7% or so in 2021, and 12% in 2022.
12% is considered fantastic in the traditional finance world.
All were running on ETH/USDT pair.
And I was testing lower balances. I was testing $1000 in each bot.
I wish I would have documented that. I didn’t start taking screenshots and documenting until I found something that was working well for me.
Wider spread won the day on Grid Bots, but 7–12% return was not enough to really make me want to dive into crypto trading.
Then I found the Martingale DCA bots.
I received an email from KuCoin, and had never heard of Martingale before. So I read about it, and decided to test them.
I thought the APR claims were ridiculous, but I had to see for myself.
After the first month, they were already netting me more than grid bots ever did, so I kept going.
The bots on Pionex offer two modes of operation :
AI Settings or Manual Configuration.
On paper and in the short term, it looks very much like the AI Settings will underperform my own Semi-Automated strategy.
Very much so.
This is why I have been manually configuring and responding to market conditions on the fly, stop/starting the bot over and over, for the last 15 months or so.
So far, what I have seen, when I do not touch my bots and leave them on AI Settings, they can end up in the red at the end of the month, they can earn as little as $130 or so (like my LTC bot did in Jan 2024 when I didn’t touch it at all).
During Dec of 2023 and Jan 2024, I completely ignored my bots, left them on a wider spread, and went about my business.
At the end of the month :
ETH = $741
BTC = $336
LTC = $132
Definately no where near $1500–2000/mo on each bot, which is what I have been getting with my manual profit capture strategy.
The highest return I have seen has been around $880 when leaving the settings alone.
However, I have NOT run AI Settings for more than 3 months consecutively.
Am I jumping the gun here?
I have not actually tested the AI settings for an entire year.
I have recently spotted some interesting claims from other users of Pionex that lead me to believe that using the default AI settings might actually only slightly underperform my own Semi-Automated strategy over a longer term.
They are claiming that AI Settings Backtesting of BTC/USDT on a Martingale bot for 1 year yielded 205% return.
Just check out these charts :
My own “See $100, take $100” strategy is yielding around 250% return.
I am doing this because — when testing the AI settings for a couple months at a time, it was looking like my return would only be 30–60% APR, IF that.
However, if the above charts are valid, then that means my own strategy was only 20% more effective than leaving the bot alone.
If that is the case, then I will stop my Semi-Automated strategy and opt for letting the “AI Settings” rock the house. I’m fine taking 20% less, if I do not have to touch it. That’s not bad!
2x in a year isn’t ever horrible. And…..
Less work, is Less Work.
This is Automated Income Lifestyle, after all, not Semi Automated Lifestyle
The Test Begins
Time to see what will happen when I invest 10k into a BTC/USDT DCA Martingale DCA bot with AI settings, and another 10k into the same bot, but playing the Semi-Automated strategy I have been testing for the last year.
Ill run both strategies as a test, trying my best to factor for variables that may occur.
Ill post the results here, and link to a new blog post when I am finished testing. (Approximately March 2025 — since I don’t plan to start until right before this year’s BTC halving)
Variables
AI Settings
The AI settings take a look at the moving averages of the market at the moment you start your bot, and will use those to create a configuration for the bot.
These settings are usually pretty safe, they are configured on the side of less risk, but can still be very profitable.
There are two modes for AI Settings:
Balanced mode, which will buy amount coins for every 1% drop, and take profit for every 1% earned (the actual percentages may be different depending on the price movement of the coin at the time you are starting your bot).
And,
Conservative mode, which will do something more like buy an amount coins for every 5% drop and take profit for every 1% earned. It will spread the risk over a greater level of price decrease.
The Balanced mode has moderate benefits and risks, while the Conservative type has lower risks and lower returns.
I am not going to run the Convservative bot.
Just not my style.
Modus
To perform this test, I will set up a bot with the default AI settings, on Balanced mode — and then make no changes.
I will Invest, start the bot, and keep it running all year.
Then, I will create another Martingale bot with the same level of investment, but I will manually configure the settings as time goes on, dynamically capturing profit manually as I have been doing for a little over a year.
About the only thing the Pionex Martingale bot does’t do — that I wish it did — automatically change its settings while its running — based on the current market profile.
That is what it’s doing when you start it up.
It is somehow looking at moving averages, creating a default setting, and then you start it.
But the bot does NOT change those settings on the fly while it’s running.
If it did, it would be THE PERFECT TRADING BOT.
So my assumption has been, to truly optimize the performance, I’d need to give it a little push.
And so I have been.
But, I’d rather not be.
May the best strategy win!
Or in this case, may the close runner up!
If it is genuinely very close, I’ll let the looser take the win, because I’m Lazy and Want to Earn Without Babysitting my Bots!
Thank you for reading!
If you enjoyed, here are three ways you can help me out:
- Drop me a follow → Automated Income Lifestyle
- Leave an appreciation and a comment if you enjoyed
- Receive an e-mail every time I post on Medium → Click Here
Until next time….
Onward and Upward Everybody!
-Chris
#automatedincome #cryptotrading #botlife #tradingbots #strategy #paralleltesting #trader #cryptotrader
