GemloomMethodology 3
Key Moments: How Automated Trading Keeps Discipline at Entries and Exits
Finding a value depression is only the start. Most of the profit and loss in small caps happens at a handful of key moments: the first buy after a coin passes review, the second buy on a pullback, the staged sells on the way up, and the exit when a stop is hit or a delisting is announced. This paper describes how Gemloom turns those moments into rules that the engine carries out around the clock.
Key points- Buying happens in at most two parts: with an allocation of 50 USDT or more, half first, and the other half only after a drop of more than 20% and a fresh red-flag check
- Selling happens in two stages: half at the first take-profit, the rest at the second; a trailing take-profit guards the gap between them
- Stop-loss, a long time without movement and delisting notices all trigger rule-based exits
Why rules
Most trading mistakes come not from bad judgment but from bad execution: not selling into strength, not admitting a loss, missing the move that happened at 3 a.m., reacting too late to news. Small caps move more, so all of these get amplified.
Gemloom writes the timing of every buy and sell into rules ahead of time and lets the engine carry them out around the clock. In the app you set your total budget, how many coins to hold at once, your risk level and your take-profit and stop-loss plan. From then on every action follows the rules and is logged with its reason.
Buying: at most two parts
Each coin's allocation is your total budget divided by the number of coins you hold at once. As soon as a coin passes review, the engine buys in your Binance account: if the allocation is under 50 USDT, it buys the full amount at once; if it is 50 USDT or more, it buys half. If the price later falls more than 20% below that first buy, the risk officer re-checks the coin for red flags on fresh data, and only if none is found does the engine buy the other half. If a red flag turns up, no second buy is made.
The reason is that calling a bottom always involves error, and bottoming coins often dip once more before they move. Keeping half in reserve lets the position be completed at a lower price and brings the average cost down, while the fresh red-flag check avoids adding money to a project that has gone wrong. Before the second buy the stop-loss leaves room for it to happen, and both the stop-loss and the take-profit levels are measured from the average cost of the two buys.
If a coin passes review while all of your slots are full, it is not lost for good. For a few days afterwards, as long as it is still on the potential board and a slot opens up in your account, the engine re-checks it against your risk level and buys it then. Under the same verdict, coins you have already bought, or already stopped out of, are not bought again, and a coin that has clearly rallied since its review is not chased on the old verdict. If a later review approves it again, the new verdict is judged on its own.
Selling: staged and trailing take-profit
On the way up, when the price reaches the first take-profit the engine sells half the position to lock in profit and recover cost; at the second take-profit it sells everything that is left. In between, a trailing take-profit is in charge: if the price makes a new high and then pulls back by a set amount, it sells the rest, giving the gain room to run without handing it all back.
There are three plans — Steady, Balanced and Moonshot — which differ in how far away the profit targets are and how much loss and pullback they tolerate; you can also set your own. Staged selling is protected in the implementation: even if one stage is too small to meet the minimum order size, the engine will not dump the whole position because of it.
Exits matter more than entries
Any of the following triggers an exit. Each exit is logged with its reason, and the public track record shows the exit reason for every signal:
| Trigger | What the engine does |
|---|---|
| Price falls through the stop-loss | Sells and closes the trade |
| Price reaches the second take-profit | Sells everything that is left |
| After the first take-profit, price pulls back to the trailing line from a new high | Sells the remainder, keeping part of the gain |
| A long time passes with no movement | Exits on schedule, freeing capital for new opportunities |
| Binance announces a delisting | Exits without waiting for the last trading day |
| You close manually or close everything | Executes immediately |
Why it has to run 24/7
Crypto never closes. Key moments often arrive while you are asleep or at work: a delisting can be announced late at night, and a stop can be blown through in minutes. The engine runs on our servers around the clock and follows prices in real time, whether or not your app is open.
Limitations
Rules cannot prevent gaps and slippage. In extreme markets the fill can be worse than the stop price, and if liquidity suddenly drops, staged sells can take longer. Rules will also sometimes sell a move that could have been held longer. We accept these costs in exchange for every trade being predictable and reviewable.
