The single check that has saved me the most money in small caps takes about ninety seconds, and it happens before I look at a chart or read one line about the project. I open the token's main DEX pool and look at how much money is actually sitting in it. That number alone kills maybe half of the candidates people send me, because the same pattern repeats over and over. A market cap in the millions resting on a pool holding a few tens of thousands of dollars, which describes a token you can buy but cannot meaningfully sell.
Take a token quoted at a five million dollar market cap with forty thousand dollars of liquidity, a combination you can find on any screener within a minute of looking. Market cap is circulating supply multiplied by the price of the last trade. Nobody deposited five million dollars anywhere. The pool is the only place real money lives, and everything that follows comes from how small that pool is relative to the headline number.
What the pool actually holds
A standard AMM pool, the kind behind most DEX pairs, holds two assets. The token itself, and a paired asset, usually ETH, SOL, or a stablecoin. A pool showing forty thousand dollars of liquidity typically holds about twenty thousand dollars of each side, and that twenty thousand of the paired asset is the entire pot available to everyone who wants out. All holders combined are competing for the same twenty grand of real money.
Now imagine you hold one percent of the supply of that five million dollar token. Your position is worth fifty thousand dollars on paper, which is more than everything in the pool put together. There is no sequence of trades that turns your tokens into fifty thousand dollars, or anything close to it. The quoted price is real only for trade sizes near zero, and your size is nowhere near zero.
Two more things are worth checking while you have the pool open. First, which pool. Liquidity fragments across DEXes and chains, and the number that matters is the deepest pool you would actually route through rather than the sum of every dusty pair that has ever existed. Second, the paired asset. A token paired against ETH or a major stablecoin exits in one hop. A token paired against the project's own governance token, or some other thin asset, means your exit routes through a second illiquid market and the price impact compounds.
The arithmetic of getting out
The useful property of AMM pools is that the math is deterministic, so you can work out your exit before you ever enter. In a constant product pool, ignoring fees, selling tokens quoted at five thousand dollars into that forty thousand dollar pool returns you roughly four thousand. You eat about twenty percent on execution, and the token price ends up down roughly a third from where it was before you clicked. That is one holder doing one mid-sized trade.
The framing I find most useful is depth. How many dollars can trade before the price moves one percent, and how many before it moves two. For a plain constant product pool, the trade size that moves the price about one percent is roughly one percent of the paired-asset side, which works out to roughly half a percent of the headline liquidity figure. On a forty thousand dollar pool that is about two hundred dollars. Two hundred dollars of selling moves the price a full percent, and anything in the thousands means you are the chart.
Concentrated liquidity pools, the Uniswap v3 style, complicate this in both directions. Liquidity providers can stack their capital tightly around the current price, so the same headline number can be much deeper near the quote, or much shallower if the liquidity sits in ranges the price has long since left. So on concentrated pools the total is even less trustworthy, and the depth figures, where a screener shows them, matter more than the TVL.
Simulate the sell before you place the buy
You do not have to trust my arithmetic, because every swap interface will run the simulation for you, and you do not need to hold the token to do it. My routine looks like this.
- Find the deepest pool on a screener and note the liquidity figure and the paired asset.
- Open the DEX or an aggregator and set up the trade in reverse, with the token on the sell side and your full intended position as the amount. The interface quotes it whether or not you hold anything. Read the price impact line before doing anything else.
- Run it again at two or three times your size, because if the trade goes well you will be trying to move a much larger dollar amount than you put in, and that bigger number is what has to fit through the door.
- Check what happened to the LP tokens. Locked or burned is what you want to see. If one wallet holds the LP, everything above sits downstream of that wallet's mood.
- Glance at daily volume against pool size. A pool that turns over a healthy multiple of itself refills from arbitrage after your trade. A dead pool stays wherever you pushed it.
The failure mode this prevents is specific and very common. You buy fifteen thousand dollars of a thin token, it triples, you are up thirty grand on paper, and then you discover that selling costs you a large slice of the position and craters the price on the way out, at which point every other holder watching the chart heads for the same door you are standing in. People lose money on winning trades this way, routinely, and it always gets described afterward as bad luck, even though it was checkable before entry.
The size rules I actually use
My baseline rule is to keep any position I might need to exit quickly under roughly one percent of the paired-asset side of the pool, which is about half a percent of the headline liquidity figure. That keeps a one-shot exit somewhere near one percent of impact. The same rule stated the other way around: I want pool liquidity at least a hundred times my position, and more than that if I expect to be selling into weakness rather than strength.
The second rule is liquidity as a fraction of market cap. Reasonably healthy small caps have historically carried DEX liquidity somewhere in the mid single digits as a percentage of their cap. When that ratio drops under roughly one or two percent, I treat the market cap as decorative and size against the pool alone. The forty thousand dollar pool under the five million dollar cap sits under one percent, and the honest position size there is whatever you can afford to have marked as unsellable for a long time.
The third rule covers the simulation. If a simulated sell at twice my intended size shows more than two or three percent impact, I cut the size until it does not, or I skip the trade. None of this protects you from the LP being pulled outright, which is why the lock check is on the list, but it removes the slower and far more common version of the same loss.
I ended up caring about this enough that liquidity sits right next to market cap on the token scorecards we build at Blockcircle, but the core check needs nothing beyond a screener tab and a swap interface. Price the exit before you place the entry, size to the pool instead of the market cap, and thin tokens lose most of their ability to surprise you on the way out.