When to Take Crypto Profits: Sell All, Half or Trail?

Sell all, sell half or hold? Follow the same profitable opportunity through every exit

Sell-all, sell-half and hold-all should be compared on the same already-profitable position, not different winning trades. The ledger below retains all 229 original complete conditional milestones and changes only the exit policy; each original opportunity is illustrated as a $1,000 account. Returns include the original 0.20% cost per side. Adding overlapping opportunity results would not create a funded portfolio equity curve.

A partial exit is a cash-flow split: sell 50% of the existing quantity at the next daily open, then the rest at the defined day-31 open. It does not mean invest another $500, halve the price gain, or create a second independent sample. For the trailing rule, the trigger is a 10% drop from the highest completed daily close; its fill is the next open, not the exact intraday stop line.

We deliberately show the largest giveback and the largest missed-upside contrasts, chosen across all complete opportunities by hold-minus-immediate net P&L. These are illustrative extremes, not representative expected profits. Their labels are assigned after the path and cannot identify in advance which trade should be sold.

Verify: All 916 paired exit rows · Original cohort and uncertainty contract

How often did waiting beat an immediate sale—and what did waiting cost?

Later exits sometimes retained more upside and sometimes gave back cash: sell-half beat sell-all in 135/229 opportunities and trailed it in 94/229; the closing trail was almost evenly split at 115 versus 114. These are the same frozen conditional milestones, not a success rate for buying new positions. The median paired sell-half advantage was $34.89 per illustrative $1,000, computed before taking the median rather than subtracting separate policy medians.

Giveback is loss of a previously marked gain; final loss is ending below starting capital; missed upside is a later policy finishing above an earlier cash exit. All three matter, but they are not interchangeable. The successful hold cases explain why always taking all profit early can be expensive, while the failure cases explain why highest median return is not sufficient risk protection.

Choose the acceptable cash exposure, maximum giveback and future decision schedule before the next alert. A model or notification can report that a condition changed, but cannot promise a fill or reverse a gap. The existing cost sensitivity, paired quarter-block uncertainty and per-coin results remain below; correlated monthly starts limit precision.

When should you take crypto profits? The numerical conclusion

The useful historical choice is exposure versus realized cash, not a perfect top: after the first +20% marked milestone, sell-half had a +29.04% final net median and sell-all +22.70% in the same 229 complete six-coin observations. Half-selling's median paired advantage was +3.49 pp, not the subtraction of separate medians. The shared endpoint is thirty days after the first possible sale, and all returns include the earlier gain against original capital.

Keeping the whole position had the largest full-sample mean and median, but 37/229 finished below original capital despite previously reaching the profit milestone. The closing-price trailing policy had 4/229 such cases. The staged policy had none in this sample; that is an observation, not a guarantee that partial selling prevents losses.

All returns below are measured against original invested capital, not additional return earned after reaching the 20% milestone. They include entry and exit costs. The primary conclusion concerns protecting existing gains versus keeping upside; it does not tell a new buyer whether the initial investment was attractive.

Who is in the sample—and who never reached a 20% profit?

The experiment begins with 432 scheduled monthly first-day purchases across BTC, ETH, SOL, BNB, XRP and DOGE. Each purchase has at most 90 daily closes to reach a 20% marked gain net of entry cost. The ledger contains 229 complete milestone outcomes, 13 reached milestones awaiting a complete exit window, 184 that did not reach the target within 90 days, and 6 still awaiting the milestone observation limit.

The 229 complete cases are therefore selected on having already become profitable. The exit comparison must not be described as a 100%-winning entry system, nor used to infer the return of all 432 initial purchases. Monthly purchases are separate equal-capital observations, not a funded portfolio or dollar-cost-averaging simulation.

Overlapping monthly entries can share the same subsequent market move. The 229 complete observations correspond to 200 unique coin/decision-date combinations. Calendar-quarter clustered intervals reduce the pretense of independence; they cannot eliminate dependence extending across quarters.

Research workflow: from raw candles to a fair policy comparison

The exit experiment asks what happens after an existing position first reaches the milestone; it must retain purchases that never qualify without assigning them a successful exit. All four rules use the same eligible monthly starts, milestone and endpoint, so differences represent exit choices under the stated assumptions, not a new winning-entry claim.

1. Freeze the official daily inputs and cutoff; validate OHLC values and continuity. Fix the six-coin scope and retain original UTC timestamps and retrospective retrieval time separately.

2. Schedule monthly first-open purchases, observe at most 90 daily closes for the first +20% marked milestone, and retain non-reachers and pending cases in the ledger.

3. Replay each preset policy on the same complete decision cohort. Decisions use closed information; fills use next-open proxies. Keep unfilled cash and proportional costs explicit. Stop all comparisons at a common endpoint.

4. Compute full-cohort mean, median, losses, paired differences and quarterly clustered uncertainty. Disclose per-coin/time-period and threshold/cost sensitivities rather than selecting only favourable groups.

5. Publish candidate rows, original inputs, exact code and file hashes. Select chart examples mechanically after the full-cohort results; the visual examples do not change which observations enter any statistic.

Which price, milestone and trailing measures actually control the exit?

Only the first +20% marked-gain milestone and each specified sale or highest-close trailing rule control the tested exits. The retained table is a shared field dictionary with the linked pump-entry study: its pump trigger, pullback, BTC SMA30 regime and volume split describe that separate entry experiment or chart context, not filters used to select profitable exits here. Daily volume and prior-price references help read the path but do not trigger a sale.

RSI, MACD, ADX, ATR, funding and open interest were not used to select trades in these experiments. We do not add them to a chart and imply they validated the outcome. Testing those combinations would require a separate rule definition, sample analysis and validation; the original results here remain unchanged.

How to read the historical price charts

We show three predetermined outcome ranks rather than choosing only attractive examples: the 10th, 50th and 90th percentile positions, index=floor((N−1)×q+0.5), in the complete cohort sorted by baseline net outcome and then observation ID. Baseline means immediate entry for the pump study and hold-all for the exit study. A middle-ranked case is not a synthetic average path; upper/lower ranks are not the best/worst possible future outcome.

Each chart preserves the daily price path and shows where each rule acted. Comparisons in the article still use the full cohort. The chart index is signal close=100 in the pump study and original purchase open=100 in the exit study. The policy-value panel uses initial cash=100 after proportional execution costs; it is not a compounded multi-trade portfolio.

BNB: holding after the milestone gave back the gain and original capital

The BNB October 6, 2025 milestone illustrates profit giveback: the original monthly buy was 1008.51 USDT; the first full or half sale used 1223.17 on October 7, but the common November 6 exit was 959.32, below the raw purchase price. The closing trail sold at 1102.57 on October 11. Read after-cost policy cash rather than assuming a previously positive mark remained secured.

This is sorted observation 24 of 229 by hold-all net outcome, with observation-ID tie-breaks. It is a lower-tail historical illustration, not a detector for a forthcoming top. Early cash sales change the exposure that remains; the trailing line cannot guarantee its own reference execution during a real gap.

Read top to bottom: actual daily price path and execution markers; relative quote turnover; then the after-cost value of each policy. The last price point is the common exit open, not that day’s later close. The table identifies every execution proxy. The +20% milestone is after entry cost; a 10% closing trail is not an intraday stop. Cash already sold stays constant.

BTC: keeping exposure retained upside beyond the first profit milestone

The BTC January 13, 2023 milestone shows the opportunity cost of selling all too early on one continuing path. The monthly buy used 16541.77 USDT; the full or first-half sale used 19930.01 on January 14; the common February 13 reference exit was 21782.37. The trailing policy also sold on that final date. Exposure kept after the milestone participated in the later rise, whereas proceeds already in cash did not.

This is sorted observation 115 of 229 by hold-all net outcome, with observation-ID tie-breaks. It is a middle-ranked actual path, not an average trajectory or proof that retaining every position is better. Compare it with the BNB giveback and complete conditional sample before choosing the trade-off.

Read top to bottom: actual daily price path and execution markers; relative quote turnover; then the after-cost value of each policy. The last price point is the common exit open, not that day’s later close. The table identifies every execution proxy. The +20% milestone is after entry cost; a 10% closing trail is not an intraday stop. Cash already sold stays constant.

DOGE: a continuing rally made the timing of partial sales matter

The DOGE February 28, 2024 milestone shows why realizing all cash at the first target can miss later upside. The original buy used 0.08956 USDT; the first sale used 0.11585 on February 29; the closing trail sold at 0.15304 on March 6; the common March 30 exit was 0.213. Partial selling preserved exposure for the final leg while realizing some earlier proceeds.

This is sorted observation 206 of 229 by hold-all net outcome, with observation-ID tie-breaks. It is an upper-tail illustration selected retrospectively, not evidence that DOGE or any altcoin should always be held. Late sale had an opportunity benefit on this path and a damaging giveback cost on other paths; the entire ledger retains both.

Read top to bottom: actual daily price path and execution markers; relative quote turnover; then the after-cost value of each policy. The last price point is the common exit open, not that day’s later close. The table identifies every execution proxy. The +20% milestone is after entry cost; a 10% closing trail is not an intraday stop. Cash already sold stays constant.

What exactly do selling all, selling half and a trailing exit mean?

We cannot sell at a daily high that was only known afterward. Each decision is made using a completed close and executed using the following daily open as a proxy. The common endpoint is the open 31 dates after the milestone candle, 30 days after the first possible sale.

The milestone tests close / [entry open × (1 + buy cost)] − 1 ≥20%. This marks the held asset after purchase cost but before a future selling charge. The observed next-open exit can be above or below 20% because of daily overshoot, gaps and the selling cost.

Which policy gave back the least—and what does giveback measure?

Here “giveback” means the highest marked portfolio value observed from the milestone through the follow-up, minus final proceeds, expressed as percentage points of original capital. Daily marks include cash already realized plus remaining units valued at that close after an estimated selling cost. It is peak-to-final giveback, not maximum peak-to-trough drawdown.

Selling everything stops later price exposure; its small residual giveback can reflect the gap from the milestone close to the next-open sale. Selling half leaves only half the original units exposed. Holding all retains the most upside and the full downside. An average across observations is not a guaranteed cap.

Is taking half off a useful compromise?

In this sample, selling half beat the immediate full sale in 135/229 (59.0%) paired observations. It did not beat it in every market. The mean paired advantage was +15.85 pp with a clustered 95% interval of +5.18 to +25.59 pp. The median paired advantage was only +3.49 pp; large runners affect the mean.

The staged policy is mathematically halfway between the immediate and hold-all proceeds for each observation under the same proportional costs. It does not discover an extra source of return. It trades part of the future exposure for realized cash. Two half-sized sell transactions do not incur twice the full-position proportional selling cost.

A common AI question is “I am up 30%; should I sell half?” This study offers a comparison framework rather than a personal instruction. Account size, concentration, future cash needs, taxes, actual spread and tolerance for loss were not observed. Partial selling can still lose original capital if the retained portion collapses far enough.

Does a trailing stop protect profits better than a fixed take-profit?

The tested 10% closing trail had a median final net result of +25.47% and 4/229 (1.7%) outcomes below original capital. It beat an immediate full sale in 115/229 (50.2%) observations. It is neither a perfect top detector nor the same mechanism as an exchange’s intraday trailing-stop order.

A path can close well below its previous high and then open even lower before this policy exits. The policy also exits after a temporary closing pullback that may later reverse. Daily OHLC cannot establish intraday ordering, stop priority, order-book capacity or whether a limit order filled.

The experiment compares one preset 10% trailing distance, not the “optimal” distance for every coin or volatility regime. RSI, funding and news-based exit filters were not fitted to these results. Testing many distances and advertising only the winner would require a separate validation design.

Does taking profit work the same way in every market period?

No. The recent subperiod is an important counterweight to the full-sample average. Positive paired differences below favour delayed or partial selling; negative differences favour immediate realization. These are descriptive time splits, not independent holdout tests.

The result can depend on market regime, coin composition and the monthly-entry cohort. A policy that benefits from unusually strong continuation in one period can give back gains in another. There is no evidence here for promising a single universal take-profit percentage.

Bitcoin versus altcoins: do the exit medians differ?

The relative medians differ by coin: BTC's trailing median (+27.35%) exceeded its half-sale median (+25.97%), while DOGE's trailing median (+15.61%) was below its sell-all median (+24.11%). Those 31 and 45 conditional milestones use the same rules and cost allowance; the contrasts challenge a universal exit rule, not forecast the best policy for either coin now.

BTC, ETH, SOL, BNB, XRP and DOGE are current surviving markets, excluding many failed or illiquid tokens. Their dated conditional samples cannot be used as a present-day token ranking or independently validated probabilities.

A larger altcoin continuation mean may simply reflect a few extraordinary rallies. The downloadable summaries include means, medians, 10th/90th percentiles, trimmed means and worst outcomes so readers do not have to rely on a single attractive statistic.

Should you take profits at 10%, 20% or 30%?

This experiment does not choose a universal 10%, 20% or 30% profit target. Higher milestones select fewer already-successful purchases: complete counts fall from 300 at 10% to 229 at 20% and 187 at 30%. A higher conditional return after a higher target does not show that waiting for it was better across all initial purchases.

The 10% and 30% thresholds were preset sensitivities alongside the primary 20% rule, not fitted after results. Each changes the qualifying cohort and when the exit decision becomes available; retain non-reachers and pending cases rather than treating them as failed sales or zero profits.

The table deliberately reports complete milestone counts. It does not compare all initial purchases under complete portfolio policies, and no “best profit target” is selected from these conditional samples.

Do fees and slippage change the comparison?

Costs reduce cash realized even after a profit milestone: the sell-all median in the same 229 complete cases falls from +22.94% at 0.10% per side to +21.96% at 0.50%. The primary comparison uses 0.20% on each buy or sell side, combining fees and slippage as an allowance rather than quoting your venue tier.

One initial cash unit buys 1/[open×(1+cost)] units, and selling receives units×open×(1−cost); cash earns zero. Half-sized sales apply costs to their actual quantities, not twice to the whole original position. Fixed order charges, tax and executed spread at size are not measured by these proportional-cost sensitivities.

Sensitivity keeps the primary decision cohort fixed, including the original +20% milestone dates in the exit study. That isolates execution cost from a change in event selection. Next-day opens are still optimistic executable-price proxies, especially after news; a daily candle cannot establish available size or fill quality.

How much confidence should you place in the historical advantage?

These intervals estimate uncertainty in paired average differences under quarterly clustering. They do not prove a causal effect, eliminate survivor selection or establish future profitability. The much smaller paired medians show why a tail-sensitive average should not be the only decision input.

Subgroup and sensitivity comparisons are disclosed together, not used to choose a winning exit rule after the fact. The preset document was written before this calculation run, but the whole analysis is retrospective and is not presented as a prospective preregistered trial.

How to use this research without trying to guess the exact top

Separate three objectives: realizing cash now, preserving upside exposure and limiting subsequent giveback. Selling all, selling part and using a conditional exit prioritize these objectives differently. Name the objective before selecting a rule; a rule cannot maximize all three for every future path.

Specify the fraction, trigger, price reference, observation frequency and next action in advance. “Sell if it feels weak” cannot be reproduced. “After a completed daily close is 10% below the highest close since my milestone, sell at the next available execution” is explicit, though still exposed to gaps and costs.

Ask an AI to distinguish realized cash from paper gains, compare the same initial capital and end date, include the outcomes that never reached the profit threshold, and identify any assumed fills. Do not let an answer turn a conditional sample of winners into a success rate for all trades. An alert can draw attention to a condition; it is not a sale or a protected profit.

Your crypto position is up 20%—how do you turn this research into an exit plan?

Start by recording your actual entry price, fees, position size and the amount of the gain you are willing to give back. The study begins only after a paper position first reached its preset +20% milestone; it says nothing about purchases that never reached that mark. Its sell-all, sell-half, trailing and hold-all outcomes are historical alternatives from that conditional point, not actions CoinNudge will execute for you.

In 229 completed milestone observations, selling half had a +29.04% median final net return versus +22.70% for selling all. Holding everything had a +31.16% median, yet 37/229 ended below original capital. A staged exit traded some remaining upside for earlier realized proceeds in this sample; it did not make a collapse impossible. The 10% trailing rule was one predeclared historical convention, not a validated best stop for every coin.

If you want to use the comparison, choose your rule before the next volatile candle: specify the milestone, fraction sold, trigger reference, how fees are paid, and what happens if a stop gaps past the intended level. Do not move the trigger after seeing the outcome and then credit the revised rule with this study's statistics.

Verify: Calculate fees and break-even before selling · Review supported Telegram market alerts

Should you sell enough crypto to recover your original investment?

Recovering the original cash outlay can reduce capital still exposed, but the remaining coins are not risk-free or free money. If your goal is cash recovery, calculate actual net sale proceeds rather than simply selling half because the position has gained 20% or 30%. The frozen sell-half experiment tests a quantity split, not a principal-recovery policy.

For original cash outlay C, current tradable quantity q, executable sale price P and proportional sell cost c, the fraction required to realize C is C/[q×P×(1−c)], before fixed charges and any tax reserve. As an arithmetic illustration only, a cost-adjusted position worth $1,200 requires selling about 83.3% to recover a $1,000 outlay—not 50%. If the fraction exceeds one, the current sale cannot recover the full outlay. Do not treat the example as an observed study return or a live quote.

After selling, track realized cash, remaining units, remaining market exposure and any tax separately. A retained token can still lose all its market value; realized proceeds also carry custody or currency risk. Tax treatment depends on jurisdiction and is not modeled in this paper. A decision based on future cash needs or concentration may favor a different fraction than the studied half-sale, and that altered policy does not inherit the original statistics.

What if crypto keeps rising after you take profits—or gaps through your exit?

A missed later rally does not by itself make a planned exit wrong; an exit trades future exposure for cash. Conversely a stop or closing trail is not a protected net-profit guarantee. The frozen daily-close trail has four below-original-capital outcomes among 229 milestones despite beginning after a marked gain, and the recent 2025–cutoff period favored immediate selling on average.

Keep the original exit record and any new entry as separate decisions. Rebuying, moving to break-even, trailing an intraday price or reallocating the proceeds to BTC were not part of this four-policy experiment. Before testing one, define its trigger, cash budget, costs, common comparison endpoint and which outcomes remain in the denominator.

If your venue order fails or price gaps below the planned reference, verify actual fills and remaining quantity before sending another order. Do not book the intended price as a completed sale. Reassess residual exposure and risk rather than changing the historical rule to match the chart. The separate fee and stop-loss studies explain those mechanics; their statistics are not interchangeable with these conditional profit milestones.

Verify: Calculate trading fees and the real break-even price · Compare crypto stop-loss rebound and further-loss paths

How can a researcher reproduce these results?

Download candle-inputs.json and decision_studies.py into one folder. Run: python decision_studies.py candle-inputs.json reproduced.json. It uses the Python standard library; the public script includes event selection, costs, exits, grouping and the seeded bootstrap. Compare reproduced.json with research.json. JSON values retain full precision; page values are rounded only for display.

The ledger includes complete, excluded, missing and pending observations rather than only winning trades. Unix times are UTC seconds; signal_day is an ISO UTC date. event_time is the earliest decision boundary after the trigger candle closed, not the retrospective data retrieval time. Entry and sale timestamps identify next-open execution proxies. This release does not establish that CoinNudge recorded those historical observations live.

Confidence intervals apply to paired mean return differences, not to medians, profit probabilities or a future trade. All observations in the same UTC decision quarter are resampled together, including different coins and overlapping monthly entries. We draw the original number of quarters with replacement 1,000 times, use Python Random seed 18092026, sorted quarter keys and linear percentile endpoints at 2.5% and 97.5%. At least 20 observations and eight quarters are required; fewer are labelled exploratory. Cross-quarter dependence and multiple descriptive comparisons remain limitations.

Every file has a SHA-256 entry in the manifest. Hashes establish file integrity, not exchange accuracy. Public research downloads are available without a paid plan. Commercial API/file products have their own coverage and entitlement; this article does not promise these experimental policies as a paid API or an automated trade service.

Verify: Full results and observations · Candidate and outcome CSV · Summary CSV · Daily inputs JSON · Daily inputs CSV · Standalone reproduction code · SHA-256 manifest

Frequently asked questions

Sources and verification

• Binance spot Kline definitionsOfficial open time, OHLC, quote volume and taker-buy fields.
• Binance historical-data documentationOfficial public-data provenance and timestamp-format context; this edition uses the public REST endpoint.
• CoinNudge frozen input acquisitionAll 13,248 daily bars and exact request URLs and retrieval timestamps.
• CoinNudge experiment and outcomesAll candidates, selected observations, costs, grouped comparisons and sensitivity.
• Historical chart data and selection contractThree mechanically selected actual paths; daily values, costs and exact execution markers.

Keep the decision context—not just the price alert.

Preview supported Telegram alerts or inspect research-data samples. No automated execution is included.