# 3 Months Running a $33 Crypto Trading Bot. Here Are the Actual Numbers.

I wrote a post a while back about deploying a trading bot with $33 and watching every single trade close red. If you missed it — yeah, every one.

So what happened next? That's what this post is about. Spoiler: the account didn't blow up. But "not blowing up" isn't exactly a flex.

All of this is with money I can afford to lose. None of it touches rent or groceries.

Where Things Stand Now

Last time it was $33 on a single pair. One strategy. One way to lose money. I've since diversified — if you can call it that at this scale.

	Before	Now
Balance	23.29 USDT	38.97 USDT (includes ~$16 added deposit)
Pairs	BTC/USDT only	BTC, SOL, ETH — three pairs
Strategies	EMA Crossover	EMA + MACD + 4h Swing
Mode	Just switched from DRY_RUN	Live, $1.0 per trade

I added MACD and a 4-hour Swing Trading strategy because the EMA Crossover on the daily chart was way too quiet. 34 trades in 37 months — that's barely once a month. I needed more signal volume, even if each individual trade is small.

Three pairs instead of one. Not because SOL and ETH are uncorrelated with BTC (they mostly are correlated), but because it gives more chances for signals to fire. Diversification here is more about opportunity than risk reduction. Honest about that.

The CircuitBreaker I built after the first disaster is still running. Daily -3%, weekly -7%, monthly -15% — three tiers of automatic shutdown. If the monthly threshold gets hit, the bot goes fully offline until I manually reset it. So far it hasn't fired. Which is either good news or means the bot isn't trading enough to trigger it.

Oh, and the balance going from 23.29 to 38.97? That's mostly me depositing more money. Not bot profits. I want to be upfront about that because it matters.

The Actual Numbers

Three months of live operation (January–March 2026).

Honestly? The bot sleeps more than it trades. A daily EMA Crossover produces maybe 1–2 signals per month. The bot checks every hour, finds nothing, logs "No crossover detected, holding," and goes back to sleep. Repeat 24 times. That's a day.

Here are the actual closed trades:

Date	PnL	Notes
2/11	-$0.002	First EMA trade. Basically zero
2/11	$0.00	Same day, second trade. Flat

...That's it. That's the list.

With the 4h Swing and MACD strategies added, I expect the frequency to pick up. But three months in, the overwhelming experience of running a live bot is waiting. During development it felt like things were constantly happening. In production? Silence. It's boring. That's probably fine.

What the Math Says

The backtest (2023–2026, 37 months) gives these projections:

Investment	Monthly Return (Optimistic)	Monthly Revenue	Annual
38.97 USDT	4.85%	$1.89	$22.68
100 USDT	4.85%	$4.85	$58.15
500 USDT	4.85%	$24.23	$290.77
2,063 USDT	4.85%	$100.00	$1,200.00

You need $2,063 to hit $100/month. With $38.97, you get $1.89. Can't even buy a coffee.

And that's the optimistic scenario — backtest returns reproducing at 100%. Realistically, 50% reproduction is generous. So cut everything in half: $0.94/month.

Compound it with full reinvestment:

Years	Balance (Optimistic)	Monthly (Optimistic)	Balance (Realistic)	Monthly (Realistic)
1	$68.77	$3.34	$51.94	$1.26
3	$214.12	$10.37	$92.26	$2.23
5	$666.73	$32.33	$163.86	$3.97

Five years of compounding in the optimistic case gets you to $32/month. Realistic: $4/month.

$33 to $100/month isn't happening. I need to accept that.

I backtested 50 strategies, celebrated a 1.30 Sharpe and 491% return, and then realized it doesn't matter what your percentage gain is when your principal is $33. A 491% return on $33 is $162. On $10,000 it's $49,100. Capital wins.

What's Next

Bot profits alone won't get me there. So I'm diversifying the income, not just the trades.

Writing about it. You're reading one of those posts right now. If this is useful to someone starting out — great. If you're setting up a bot, you'll need an exchange account, and picking one with low fees saves real money over time.

About Exchanges

I run my bot on Bitget. But if I were starting today, I'd probably go with MEXC. The reason is simple: zero maker fees on spot.

Exchange	Maker Fee	Available Worldwide	ccxt Support
MEXC	0% (Spot)	Yes	Yes
Bitget	0.1%	Yes	Yes
OKX	0.08%	Yes	Yes
Bybit	0.1%	Restricted (2024+)	Yes

When your bot trades automatically, the fees compound. Even 0.1% per trade, 10 trades a month, adds up to ~12% annual drag. Zero fees eliminates that entirely.

You can create an account at https://www.mexc.com/acquisition/custom-sign-up?shareCode=mexc-3qBzR. It's free, takes a couple minutes. Generate your API key, put it in your .env, and make sure withdrawal permissions are OFF. If you want to try running a bot yourself, https://www.mexc.com/acquisition/custom-sign-up?shareCode=mexc-3qBzR.

Signal service. I've already built the Telegram signal module — free channel and premium channel, with the bot posting trade signals automatically. Premium would be $29–49/month. Even a handful of subscribers gets me to $100. Still need to set up the actual Telegram bot, though. Soon.

Keep growing the bot. Reinvest everything, run three pairs × three strategies, let compound interest do its slow thing. Starting from $33 means it takes forever, but there's no reason to stop.

The Takeaway

Three months of P&L: roughly break-even. Not profitable.

But also not blown up. The CircuitBreaker is doing its job. The system — 50 backtested strategies, three live pairs, automatic risk management, DRY_RUN mode — works. The code is solid. The returns just aren't there yet because the capital isn't there.

$33 to $100/month was never realistic. That's a fact. But having a bot that runs autonomously, checks the market every hour, and doesn't blow up your account while you sleep? That's not nothing.

Not quitting. (Second time saying that.)

This post is based on personal experience. Not financial advice. Always use money you can afford to lose. Backtest results reflect historical data and don't guarantee future returns. This post contains affiliate links.