Independent evidence review. “Does MoonBot work?” is too broad to answer honestly with one green or red label. MoonBot is trading software with many manual and automated capabilities. BibaMoney tested two narrower claims: whether four MoonBot-style strategy replicas showed a live-ready edge, and whether a frozen cohort selected from public MoonBot trader statistics could be copied as a repeatable net-positive strategy.
This is not a finding that MoonBot software is fraudulent, unsafe in every configuration or incapable of placing orders. It is also not a tick-faithful reproduction of the vendor's engine. The evidence rejects two specific automation claims at this cutoff.
MoonBot review: three scopes that must stay separate
| Scope | What was observed | What the result means |
|---|---|---|
| MoonBot terminal | Official documentation, API permissions and stated product features | Software capability and live-access exposure, not strategy profitability |
| BibaMoney replica | Four operator-configured economic proxies on 5-minute candles | A directional test of these rules, not the sub-second MoonBot engine |
| Frozen copy cohort | Eight trader IDs selected once, then followed forward | A test of one selection method, not every MoonBot user |
What MoonBot officially offers
The official download page describes a portable Windows terminal that can ease or automate crypto trading and offers an emulator path before real execution. The official FAQ describes tick charts, grids, signal automation, manual and automated strategies, spot and futures support, and additional modules.
For real orders, MoonBot connects to an exchange. Its API-key documentation requires trading permission and says the private signing key is stored locally in encrypted form. Local key storage is a meaningful security property, but trading permission still creates capital risk: software or a strategy can lose funds without withdrawing them.
The official pages also contain exchange referral links, trial benefits and cashback offers. Those are disclosed commercial relationships, not evidence of fraud and not evidence of strategy edge. BibaMoney uses no reviewed-entity affiliate link in this article.
What the internal MoonBot Replica board tested
The board at /app/admin/moonbot translates four operator-configured MoonBot-style setups into reproducible economic rules:
- moontop_grid and moonmeme_grid: resting dip-buy grids based on the economic idea of MoonShot;
- btc_longs_fast: an EMA-style dip-long proxy;
- btc_shorts_fast: a short-horizon pump-short proxy.
MoonBot's own MoonShot explanation says the strategy keeps orders below market to buy a fast downward move and can freeze part of the deposit in open orders or positions. Our model preserves that economic mechanism, marks unclosed positions to market and charges 5 basis points per side.
The important limitation is resolution. MoonBot observes ticks and sub-second conditions; the replica uses 5-minute candles. We cannot know the tick path inside a candle, reproduce millisecond trade density or claim exact fill parity. No Gate.io rebate is assumed. These choices make the study reproducible and conservative, but not vendor-identical.
Replica results: three negative, one paper candidate
The frozen publication snapshot was generated from run 54, last updated at 05:30 UTC on 30 July. Results use spot leverage 1×, a rolling $100 order and both-side fees. “Net” below is research P&L for that unit contract, not a forecast for a user account.
| 5m proxy | Trades / win rate | Current net | 324-config held-out result |
|---|---|---|---|
| moontop_grid | 136 / 43.4% | −$57.55 · 9 frozen | −$14.27 · IMPROVED_STILL_NEGATIVE |
| moonmeme_grid | 167 / 49.7% | −$34.62 · 8 frozen | −$51.29 · IMPROVED_STILL_NEGATIVE |
| btc_longs_fast | 39 / 35.9% | −$0.24 | −$0.42 vs −$0.29 baseline · OVERFIT |
| btc_shorts_fast | 122 / 50.8% | +$0.53 · 2 frozen | +$0.54 vs −$1.78 baseline · PAPER CANDIDATE |
The BTC-short arm is the honest nuance. It improved on the held-out tail, so calling all four arms failures would be wrong. But +$0.54 after a 324-configuration search is too small and too selected to authorize live capital, especially when the simulation cannot reproduce tick execution. It needs a locked configuration and a sealed forward paper record.
MoonBot copy-trader evidence: why the headline changed
MoonBot's official statistics FAQ says its public statistics can show terminal IDs, trade fields, daily leaders and strategy popularity. It also says strategy information is collected only from users who allow it. That makes the feed useful telemetry, but not a complete population or a broker-statement audit.
To reduce leaderboard hindsight, BibaMoney froze eight trader IDs on 25 June and did not reselect them after seeing future results. At the publication cutoff:
- 35 forward days and 24,090 source trade records;
- 2 / 8 holds and 6 / 8 fails under the daily-cluster gate;
- hold rate 25.0%, with a wide Wilson 95% interval of 7.1%–59.1%;
- six streams have a positive source-percentage sum, but only two pass the statistical gate.
The old view effectively treated every positive cumulative percentage sum as a hold and therefore suggested 6/8. The corrected method groups trades by UTC day because thousands of intraday trades from one source are not thousands of independent experiments.
Why +1,425.11% is not an account return
The eight streams currently sum to +1,425.11% after the research fee adjustment. That number is preserved for reconciliation, but it is not ROI. It adds percentages across different traders and overlapping trades without position sizes, free collateral, concurrency, compounding or a shared account ledger.
The tail shows why this distinction matters. One failed stream displays a 95.9% trade win rate yet sums to −272.36%, with an average losing trade of 16.87%. Another displays a 97.9% win rate yet sums to −107.19%, with an average losing trade of 22.19%. Many small wins can coexist with rare losses large enough to dominate the book.
Timestamp meaning is also unresolved: our current diagnostic cannot reliably map the sampled source timestamps to entry versus exit candles. Without that semantic contract, exchange fills and measured copy slippage cannot be reconciled. The verdict therefore concerns repeatability of the selection signal, not executable portfolio ROI.
Does this establish fraud or a scam? No.
No such finding follows from this evidence. The official documentation substantiates that MoonBot is trading software with emulator and live exchange functions. Our result says that four tested proxies do not yet demonstrate a robust live edge and that one frozen public-statistics cohort did not reproduce the copying thesis under its declared gate.
Software capability, strategy quality, user behavior, exchange execution and marketing claims are separate questions. A negative strategy result cannot adjudicate all of them.
STOP / GO decision and budget
| Lane | State | Budget |
|---|---|---|
| Live MoonBot execution or automatic copying | STOP | $0 |
| Buying strategy packs, signals or validation capital | STOP | $0 |
| Affiliate promotion of MoonBot or linked exchanges | STOP | $0 |
| Read-only evidence refresh and deterministic reproduction | GO | Existing infrastructure only |
| Locked BTC-short paper candidate and strategy-audit research | GO, paper only | $0 incremental infrastructure |
What evidence would change the decision?
- Tick or trade-level replay of an exact MoonBot configuration, with exchange fee tier, rebate and deterministic fill parity.
- A locked BTC-short candidate that remains positive in a sealed forward paper record.
- Defined source entry/exit timestamps and immutable raw records for the copy cohort.
- A capital-constrained replay with concurrent exposure, fees, slippage and available collateral.
- Reconciliation against representative exchange statements or broker-controlled read-only data.
- A newly preregistered cohort that passes the same daily-cluster and tail-risk gates.
Editorial policy and right of reply
This review is not paid placement and contains no MoonBot or exchange affiliate link. MoonBot's operators, strategy authors and data providers may submit dated primary evidence, a correction or a response. Material corrections will be logged and the verdict may be upgraded, narrowed or retracted when evidence changes.
Read the FinEdg testing methodology and the related PHOENIX, ZigZag, MRS and ZZBoba evidence review. This material is research, not financial advice or an invitation to trade.