Last Updated: 2026-10-27
A coin is not an edge. A useful watchlist identifies a particular instrument, venue and observation window, then asks whether your proposed rule can be evaluated with the available data. This guide replaces a purported ten-coin ranking with a repeatable selection exercise. It supplies no measured 2026 ranking, profitable setup or recommendation to trade.
Define the experiment before choosing the coin
Write one sentence such as: "I will study completed five-minute bars for this spot pair between 13:00 and 15:00 UTC, recording every occurrence of this precisely defined rule." Specify the rule separately; a chart that looks clean is not a definition. Record whether timestamps mean bar start, bar close or receipt time.
Choose candidates you can observe consistently. BTC, ETH or any other ticker is only a label until the venue, quote asset and product are fixed. A spot pair and a perpetual on the same coin have different cash flows and risks. Derivatives access is not a prerequisite for learning market data, and a reachable website does not establish eligibility.
Use a blank comparison sheet rather than copying a list of supposed winners:
Scroll sideways to see every column. Keyboard: focus the table region and use the arrow keys.
| Field | Record for each candidate |
|---|---|
| Instrument | Venue, symbol, spot/derivative, base and quote assets |
| Observation scope | UTC window, dates, sampling interval, missing periods |
| Range | Defined calculation on completed bars, not an unexplained "typical" percentage |
| Executable depth | Buy and sell estimates for the same intended quantity |
| Costs | Account-specific fees, assumed fill method, financing if applicable |
| Exceptions | Halt, stale book, missing candle, delisting or incomplete history |
| Decision | Eligible for this study, insufficient evidence, or excluded with reason |
Keep exclusions in the sheet. Removing delisted pairs or bad sessions after seeing results makes the surviving candidates look better than the original choice set.
Compare range without mistaking it for profit
For a synthetic session with open 100, high 103 and low 99, the high-low range divided by the open is 4%. A trader could not necessarily buy the low and sell the high: their order, timing and sequence matter. The range describes the observed session; it does not forecast the next one or measure an achievable return.
Record the same range definition for every candidate and report several sessions, including quiet and disrupted ones. If one market has missing bars, mark the comparison incomplete. A volume spike can justify investigation; it does not establish a profitable signal. Use turnover in the same units and distinguish exchange-reported volume from executable resting orders.
Let costs challenge the apparent opportunity
This invented example uses 100 units bought at an actual simulated fill of $100 and sold at $100.20. Gross gain is $20. With a hypothetical 0.05% fee on each filled leg, entry costs $5 and exit costs $5.01, leaving $9.99 before other costs.
Now keep the same intended exit but assume only $100.05 is executable. Gross gain falls to $5; fees are $5 and $5.0025, so the result is a $5.0025 loss. A fifteen-cent difference in the exit fill changed the conclusion. The example does not model queue position, market impact, tax or financing.
Do not subtract the spread a second time if your entry and exit fills already incorporate crossing the book. If you instead start from mid-prices, explicitly model bid/ask execution. In either case, use filled turnover rather than multiplying a fee rate by the number of trades and calling that a percentage of the account.
For example, twenty round trips at $1,000 per leg produce $40,000 of filled turnover. At 0.05% per leg the fees total $20. That is 0.05% of turnover, or 4% of a separate $500 reference balance. Neither denominator should be hidden.
Treat stops and missing data as separate cases
A stop trigger is not an execution guarantee. A triggered market order may receive a worse fill; a price-limited exit may remain unfilled. Binance's stop-order explanation explicitly distinguishes trigger selection from execution and notes that stops cannot guarantee avoidance of liquidation. That product-specific explanation is not a claim that another venue behaves identically.
In a paper log, retain "triggered but unfilled", "partial fill", "no current quote" and "unknown order state" instead of forcing every row into a win or loss. Never substitute the next convenient candle close for an unavailable execution price without labeling the assumption.
If the studied product is a perpetual, use its actual payment schedule and observed rates. As arithmetic only, a constant 0.05% payment every eight hours for thirty days is 90 payments, or 4.5% of unchanged notional. It is not 5.5%, not a forecast, and not a continuous hourly charge. Rates, notional and eligibility at each settlement can change.
Decide what the evidence supports
Freeze the selection rule before evaluating a later period. Keep all qualifying observations, actual gaps and the cost assumptions. Compare results under less favorable fills; record how much of any apparent gain depends on a handful of observations. A small profitable sample is not a license to increase leverage or move to live orders.
The useful outcome may be "this feed cannot support the test" or "the apparent move does not cover plausible costs." There is no universal minimum account balance, number of trades or required charting subscription that proves readiness. A spreadsheet and public observations can answer the first questions without funding an account.
Use the altcoin execution worksheet for order-book arithmetic and the chart setup guide for separating study views from execution.
Educational research workflow, not financial advice. Trading can lose capital. Examples are synthetic, not personal trades or measured platform performance. The site may receive referral commissions on other pages; this guide contains no affiliate signup recommendation.