I rebuilt a viral "3,888% return" TradingView strategy — it landed almost exactly at coin-flip odds

A trading YouTube channel (Trading with DaviddTech)
posted a video claiming a "strange 63% win rate" TradingView indicator, turned into a full
strategy (three indicators stacked together, nicknamed "Manergy Cloud" in the video), backtested
on BTCUSDT at 1-hour bars with a claimed 3,888% total return, 62.8% win rate, profit factor
2.1, over 237 trades — using 10x leverage.

I wanted to actually test this, not just read the claim. This post walks through what happened
when I tried — including the part where the video's own central indicator turned out not to
actually exist anywhere obtainable.

What the video claims

• Strategy name: "Manergy Cloud," tested on BTCUSDT, 1-hour timeframe, 10x leverage.
• Entry indicators: an "EWMA" (the video's own term — stated as "Entropy Weighted Moving Average," not the standard exponentially-weighted moving average) filters for tradeable vs. choppy conditions; a public Range Filter (by Donovan Wool) sets trend direction; an "LMO" (Lead-Lag Moving Oscillator) confirms entries.
• Risk: ATR bands (3x multiple) for the stop, fixed 1:2 reward:risk.
• Claimed results: 3,888% total P&L, profit factor 2.1, 62.8% win rate (149W/88L over 237 trades), 21% max drawdown, longest losing streak 6 trades, longest winning streak 16.

Treat every number above as a claim from the video, not an independently verified fact.

The reproducibility problem — this is the actual headline finding

Before writing any code, I checked the video's own comment section. Multiple top comments (6+
likes each) report:

"The EWMA Indikator is not on Tradingview or in your website"
"Why post a video to copy a strategy but not provide the indicators?"

The video's description pushes a Skool community, a "strategyfactory.ai" resource link, and a
pinned comment offering "all my TradingView strategies & tools... for FREE. LIMITED SPACES!" —
classic lead-gen structure. Neither the "EWMA" nor the "LMO" indicator used in the video is a
standard, publicly-documented technique with a known formula, and per the video's own audience,
neither is actually available anywhere to check.

This matters more than the headline return number. It means nobody watching that video can
actually verify the 3,888%/62.8% claim — not because the underlying idea is necessarily fake,
but because the specific indicators driving entries were never published. That's worth knowing
before anyone trades real money on the strength of that number.

What I actually tested

Given the reproducibility gap, I built the closest honest equivalent I could, being explicit about
what's faithful and what's reconstructed:

I also substituted instruments: MES, MNQ, SPY, and QQQ (unlevered), instead of BTCUSDT at 10x
leverage — both because this pipeline focuses on US equity/futures instruments, and because
crypto's 24/7 session structure and leverage assumptions aren't directly comparable to what most
readers of this blog are actually trading.

Bottom line on scope: this backtest tests my reconstruction of the described strategy shape, not
a verified port of the video's actual strategy. That's a materially weaker claim than
"replicated his results" — and an important distinction, not a footnote.

The signal, in Python

# strategy_logic.py (excerpt)
class ManergyCloudSignal:
    """
    Entry confirmed when: regime filter is 'green' (trending, not choppy)
    AND Range Filter direction flips AND LMO green/red lines confirm the
    same side. Risk: ATR-band stop (3x ATR), fixed 1:2 reward:risk.
    """
    def signals(self):
        long_flip = self.df["rf_trend_up"] & ~self.df["rf_trend_up"].shift(1).fillna(False)
        short_flip = self.df["rf_trend_down"] & ~self.df["rf_trend_down"].shift(1).fillna(False)

        long_confirm = long_flip & (self.df["lmo_green"] > 0) & (self.df["lmo_red"] < 0) & self.df["regime_green"]
        short_confirm = short_flip & (self.df["lmo_green"] < 0) & (self.df["lmo_red"] > 0) & self.df["regime_green"]
        # ... ATR-band stop distance attached per signal, entry filled at next bar's open

Enter fullscreen mode Exit fullscreen mode

The Range Filter is a faithful port of Donovan Wool's published stateful smoothing-band algorithm
(a var-style filter line that only moves when price closes outside its current band, direction
flips tracked via consecutive up/down move counters). The regime filter is a Kaufman efficiency
ratio: net directional displacement over a window, divided by total absolute movement in that
window — high efficiency means trending, low means choppy.

Backtester is fully event-driven and causal: entries fill at the next bar's open after a signal
(never the signal bar's own close), stop-vs-target resolved bar-by-bar using only that bar's real
High/Low (stop wins on same-bar ambiguity), commission ($0.85/side) and slippage (1 tick/side)
applied on both entry and exit.

The same logic in Pine Script v6

//@version=6
strategy("Manergy Cloud-Style (Range Filter + Reconstructed LMO/Regime)", overlay=true)

// Range Filter (Donovan Wool, faithful)
avgRng = ta.ema(math.abs(close - close[1]), rfLen)
smoothRng = ta.ema(avgRng, rfLen * 2 - 1) * rfMult
var float rfFilt = na
rfFilt := na(rfFilt[1]) ? close :
     close > rfFilt[1] ? math.max(rfFilt[1], close - smoothRng) : math.min(rfFilt[1], close + smoothRng)

// Reconstructed regime filter (Kaufman efficiency ratio, stands in for the
// video's unobtainable "EWMA" indicator)
netMove = math.abs(close - close[entropyWindow])
totalMove = math.sum(math.abs(close - close[1]), entropyWindow)
efficiency = totalMove == 0 ? 0.0 : netMove / totalMove
regimeGreen = efficiency > (1 - entropyThreshold)

// Entry + ATR-band stop / 1:2 R:R target
if longEntry
    stopPrice = close - atr[1] * atrMult
    targetPrice = close + atr[1] * atrMult * rr
    strategy.entry("Long", strategy.long)
    strategy.exit("Long Exit", from_entry="Long", stop=stopPrice, limit=targetPrice)

Enter fullscreen mode Exit fullscreen mode

Unlike the earlier CFTC-based post on this blog, this strategy is fully computable bar-by-bar from
OHLCV alone — no external weekly data feed needed — so this is a real strategy() script, not
just a signal overlay. Same reconstruction caveat applies: the LMO and regime-filter portions are
approximations of the video's description, not verified ports of its actual formulas.

Backtest results

Hourly bars, real historical data: MES/MNQ from ~3.25 years of Tiger continuous-futures data
(Apr 2023–Jul 2026), SPY/QQQ from ~1 year of Tiger hourly bars (Jul 2025–Jul 2026 — Tiger's
hourly-history API appears capped around a year, a real data-window limitation worth flagging
rather than hiding).

All trade counts (310–3,687) are comfortably above the ~50-trade threshold I use to trust a result
at all — MES and MNQ especially so.

The actual finding: win rates cluster tightly at 33–35% — almost exactly the mathematical
breakeven point for a fixed 1:2 reward:risk setup (1 ÷ (1+2) = 33.3%). Before costs, this strategy
lands right at coin-flip-adjusted-for-payout. After realistic commission and slippage, three of
four instruments go net negative. MNQ's apparent "win" (+$11,239 net) comes with a -$111,396 max
drawdown — over 9x the eventual profit — which is nowhere close to a usable risk-adjusted result
regardless of the headline sign.

The ~65% stop-out rate across all four instruments is consistent with a strategy sitting almost
exactly at its own breakeven line: neither a clear edge nor a clear failure at the entry-logic
level, just costs and drawdown doing what they do to a coin flip run thousands of times.

Honest caveats

• This tests a reconstruction, not the original strategy. The LMO and "EWMA" components are approximations built from the video's description, not verified ports of the creator's actual (never-published) formulas. A genuinely faithful test isn't possible with what's public.
• Instrument and leverage mismatch with the source video. Tested on unlevered MES/MNQ/SPY/QQQ hourly bars, not BTCUSDT at 10x leverage — a fundamentally different risk profile than the video's claimed 3,888% figure, which should be read as a leveraged-crypto number, not comparable to an unlevered-equity-index number.
• SPY/QQQ only got ~1 year of hourly data (a Tiger API history limit for hourly bars, not a choice) — meaningfully less than MES/MNQ's 3.25 years. Their trade counts (310, 315) still clear the reliability floor, but with a shorter, more recent sample.
• Costs are estimates ($0.85/side commission, 1 tick slippage/side), not a confirmed real fee schedule.
• No partial exits or trailing stops — a fixed stop and fixed 1:2 target only, matching what the video described, though a real trader might manage the position differently.

Bottom line

The video's headline claim (3,888% return, 62.8% win rate) can't actually be independently
verified — its central indicators aren't publicly available, a fact confirmed by the video's own
audience in the comments, not just my own search. When I rebuilt the described strategy shape
using faithful public components (Range Filter, ATR bands) plus honest reconstructions for the
unavailable pieces, on real MES/MNQ/SPY/QQQ hourly data, the result landed almost exactly at the
mathematical breakeven win rate for its own reward:risk ratio — and lost money after costs on
three of four instruments tested, with the fourth "winning" instrument carrying a drawdown far
larger than its eventual profit.

This doesn't prove the original video's claim is false — it's possible the real, unpublished
indicators genuinely do something meaningfully different from my reconstruction. But it does mean
nobody can currently verify that claim from the outside, and the closest honest attempt at
rebuilding it didn't show an edge.

Full code and trade logs for this post are private for now — ask if you want them.