Every week there is a new thread. Some guy set up a bot on a Mac Mini. Maybe a Raspberry Pi if he is feeling supersmart. He found an arb. Polymarket, DEX, CEX, whatever. The screenshots look incredible for two weeks. Then the money is gone and the thread goes quiet.
I am writing this because I have been on the other side of these trades and I am tired of watching people learn this with their savings.
1. polymarket: you do not have alpha
A few months ago, me and a friend of mine, the former COO of a hedge fund @yupitstobes, sat down and tried to build Fully delta-neutral Polymarket strategies. Spread capture. The real thing.
We spent weeks on it. Built the models. Got the order book depth data that the public Polymarket API does not give you. And what we basically found some nasty stuff.
There are not enough gamblers. The liquidity sitting on Polymarket books is overwhelmingly maker flow. Sophisticated flow. Arbs, market makers, and sharp money pouring in by the thousands. The dumb money that your strategy needs to exist? It is not there. Not at scale.
The only places with enough flow are the top liquidity markets. BTC and ETH prediction markets, but there you are trading against TradFi desks who already had the infrastructure and just plugged it in. Sports markets, you are up against bookmakers who have been doing this for decades. There are cracks to exploit, but the edge is razor thin and shrinking daily.
Our main focus was political events. Presidential markets. What we discovered, once we actually had depth data, was that the majority of resting orders were from other people trying to do exactly what we were doing. Everyone was a market maker. Nobody was the fish.
If you think your laptop bot found an edge on Polymarket, you almost certainly did not. The people filling your orders have more data, more capital, and faster pipes. Whatever profit you see early on is noise. It will mean-revert into a loss. Your operations will not scale. You will find this out the hard way or you will read this and save yourself the tuition.
2. you will lose more in tokens than you think
Most bot operators obsess over trade PnL and ignore everything around it. The gas on failed transactions? Adds up fast, spread you think you captured? After fees, slippage, and bad trades it is a fraction of what your backtest told you.
Backtests are liars. They do not know that your order would not have been filled at that price because someone faster already took it. They do not model the market impact of your own orders. And they absolutely do not model the three days your Mac Mini was down because it overheated next to your router.
The gap between backtest PnL and live PnL is where retail traders go to die. It is not a small gap. It is a canyon.
3. fees will eat you alive
Even if you have real alpha, which you probably do not, fees will destroy you if you are not obsessively managing them.
On DEXs: gas on every tx, swap fees on every trade, slippage on every execution. On CEXs: maker/taker fees, withdrawal fees, and the invisible cost of being last in queue while everyone with colocation eats your lunch.
The market is designed to extract money from people who do not understand the full cost structure of their own strategy. If you cannot list every fee your bot pays per trade from memory, you are the one being extracted from.
4. if everyone is a shark, no one eats
There are firms whose entire existence is eating small fish. Dedicated teams. Proprietary infra. Direct exchange connections. Years of data. Their only job every single day is finding and extracting the exact edge you think you just discovered on a Sunday afternoon.
When you drop your Mac Mini bot into a market, you are not trading against other guys with Mac Minis. You are trading against them. And when every participant is sophisticated, when the pool is nothing but sharks, margins compress to zero. There is nobody left to take money from.
This is what we found on Polymarket. This is what happens on mature DEX pairs. This is what will happen to your bot. The question is not if. It is when.
5. some light at the end of the tunnel
Not everything is hopeless. Some strategies can scale. You can actually make money. Your alpha might dilute over time, but if you found something real, you can make your buck before it does.
The difference between strategies that print and strategies that blow up is not the idea. It is never the idea. It is the execution.
Strategies that work are built on real infrastructure, not on a laptop. They account for every cost before the first trade. They hedge inventory risk instead of praying. They do not carry directional exposure overnight and hope. And the people running them know exactly who is on the other side of every trade they take.
If you are serious about this, invest in your brain first. Understand the microstructure. Know who you are trading against. And be honest with yourself about whether your edge is real or whether you are just the next fish swimming into a pool full of sharks who have been doing this longer than you have been in crypto.