Indicator Lab · 01 · EMA
Analyse EMA rules on BTCUSDC over the last 30 days and understand every step at the same candle: from the weighted average and BUY/SELL to execution, costs and results.
Examine a clearly defined crossover rule over the latest 30 days. Chart, BUY/SELL action, costs and journal belong to the same analysis. Explanations accompany the application; below, inspect the calculation for the same candle.
One tool · one dataset · the same assumptions
BTCUSDC · 1-hour candles · 30 days
Shared market snapshot: 11 Oct 2026, 15:52 UTC · Bootstrap snapshot · refresh scheduled · Age when opened: 3,136.0 min
Evaluation window: 11 Sept 2026, 15:00 — 11 Oct 2026, 15:00 UTC (end exclusive) · 720 confirmed hourly candles
3000 earlier candles supply EMA warm-up only; they do not count toward the 30-day performance. A page visit can refresh a stale server cache in the background for later visitors. Your opened snapshot stays fixed until you fetch fresh data yourself.
Fetching fresh updates the snapshot. Analysis still uses completed hourly candles; there is no tick stream.
Number of candles. Longer smooths more and reacts more slowly.
The chart and trading rule use the same two EMA series. The fast period must be shorter than the slow period. A longer EMA smooths more and reacts more slowly. The optional SMA weights the latest N closes equally and does not alter the EMA rule.
Spot, long or cash. EMA 10 crosses strictly above EMA 30 at a confirmed candle close: buy. A cross strictly below: sell the entire position. Fill after 1 candle(s) at the open plus adverse slippage. No leverage, additions or price stop. A delayed order remains scheduled until execution.
Start in cash; an existing above-state does not open a position. At each new entry, the chosen fraction of available cash includes the buy fee. Fractional quantities are allowed; exchange filters, financing and market impact are not modeled.
One adverse markup/markdown; do not add spread twice.
1 = next open; never backdated to the signal close.
Descriptive analysis of this 30-day snapshot. Parameter comparisons on the same history are not an independent out-of-sample test.
Public market snapshot
BTCUSDC · BTCUSDC · 1h · last 30 days
The chart window limits the display only. The rule analysis uses all disclosed candles in the 30-day evaluation window.
Practise the rule: choose BUY, SELL, HOLD or WAIT, check the explanation, then reveal the next candle.
A SELL crossover can result in WAIT when there is no long position to sell. The action accounts for the position and queued executions; only dots mark actual trades.
HOLD · Hold position
11 Oct 2026, 14:59 UTC · confirmed close
No new confirmed crossover at this close. The existing position stays open. Remaining above does not constitute a new buy.
Signal at this close
No new signal
Actually filled at this open
No fill at this candle.
Historical simulation of this EMA rule with your parameters. BUY/SELL describes its action, not personal trading advice or a profit forecast.
Gross P&L 703.11 − fees 281.15 − Slippage 140.57 = net P&L 281.39
Gross uses actual traded quantity and the reference prices of filled orders. It is not a separate zero-cost backtest. A position still open at the end is marked at the close; no sale or exit cost is invented.
Value range: 9,556 — 11,073
Cash baseline: 10,000.00 · Buy and hold starts at the first evaluation open with the same capital, allocation and cost model.
Open position: 0.123791 · unrealized net P&L including paid entry fee -31.11.
| Experiment | Net | Drawdown | Trades |
|---|---|---|---|
| A · EMA 10/30 | 2.81 % | 8.02 % | 13 |
| B · EMA 15/30 | 0.84 % | 9.96 % | 10 |
| Only fee +0.10 percentage points | 0.08 % | 9.84 % | 13 |
| Only execution +1 candle later | -1.47 % | 11.51 % | 13 |
Data, period and all unnamed assumptions remain identical. Quantity and cash are recalculated when costs change. No automatic parameter selection; repeated experimentation is not independent evidence.
Wins / losses / breakeven: 4 / 9 / 0. Completed-trade win rate: 30.77 %. Not an estimate of the next trade’s win probability.
A journal click opens the history available through that point. Later trades disappear in replay.
| Candle | Side | Quantity | Open → fill | Fee | Cash after | Replay |
|---|---|---|---|---|---|---|
| 3054 | Buy | 0.129226 | 77,268.02 → 77,306.65 | 9.99 | 0.00 | |
| 3057 | Sell | 0.129226 | 76,700.62 → 76,662.27 | 9.91 | 9,896.83 | |
| 3062 | Buy | 0.127392 | 77,571.37 → 77,610.16 | 9.89 | 0.00 | |
| 3088 | Sell | 0.127392 | 77,614.44 → 77,575.63 | 9.88 | 9,872.66 | |
| 3134 | Buy | 0.129060 | 76,382.09 → 76,420.28 | 9.86 | 0.00 | |
| 3209 | Sell | 0.129060 | 80,295.80 → 80,255.65 | 10.36 | 10,347.44 | |
| 3219 | Buy | 0.127058 | 81,316.89 → 81,357.55 | 10.34 | 0.00 | |
| 3288 | Sell | 0.127058 | 85,788.80 → 85,745.91 | 10.89 | 10,883.78 | |
| 3333 | Buy | 0.128283 | 84,714.68 → 84,757.04 | 10.87 | 0.00 | |
| 3336 | Sell | 0.128283 | 83,909.60 → 83,867.65 | 10.76 | 10,748.05 | |
| 3369 | Buy | 0.127349 | 84,272.02 → 84,314.16 | 10.74 | 0.00 | |
| 3395 | Sell | 0.127349 | 84,122.52 → 84,080.46 | 10.71 | 10,696.85 | |
| 3427 | Buy | 0.127380 | 83,849.81 → 83,891.73 | 10.69 | 0.00 | |
| 3435 | Sell | 0.127380 | 83,036.46 → 82,994.94 | 10.57 | 10,561.36 | |
| 3454 | Buy | 0.125713 | 83,885.71 → 83,927.65 | 10.55 | 0.00 | |
| 3467 | Sell | 0.125713 | 83,448.53 → 83,406.81 | 10.49 | 10,474.84 | |
| 3471 | Buy | 0.124429 | 84,057.47 → 84,099.50 | 10.46 | 0.00 | |
| 3511 | Sell | 0.124429 | 84,438.78 → 84,396.56 | 10.50 | 10,490.84 | |
| 3533 | Buy | 0.123332 | 84,934.01 → 84,976.48 | 10.48 | 0.00 | |
| 3535 | Sell | 0.123332 | 84,744.79 → 84,702.42 | 10.45 | 10,436.11 | |
| 3544 | Buy | 0.122731 | 84,905.04 → 84,947.49 | 10.43 | 0.00 | |
| 3580 | Sell | 0.122731 | 85,430.19 → 85,387.47 | 10.48 | 10,469.21 | |
| 3584 | Buy | 0.121631 | 85,944.84 → 85,987.81 | 10.46 | 0.00 | |
| 3589 | Sell | 0.121631 | 85,534.17 → 85,491.40 | 10.40 | 10,387.97 | |
| 3596 | Buy | 0.120631 | 85,984.50 → 86,027.49 | 10.38 | 0.00 | |
| 3606 | Sell | 0.120631 | 85,616.28 → 85,573.47 | 10.32 | 10,312.50 | |
| 3670 | Buy | 0.123791 | 83,181.06 → 83,222.65 | 10.30 | 0.00 |
| # | Entry → exit | Gross | Fees | Slippage | Net | Candles |
|---|---|---|---|---|---|---|
| 1 | 3054 → 3057 | -73.32 | 19.90 | 9.95 | -103.17 | 3 |
| 2 | 3062 → 3088 | 5.49 | 19.77 | 9.88 | -24.17 | 26 |
| 3 | 3134 → 3209 | 505.10 | 20.22 | 10.11 | 474.77 | 75 |
| 4 | 3219 → 3288 | 568.19 | 21.23 | 10.62 | 536.34 | 69 |
| 5 | 3333 → 3336 | -103.28 | 21.63 | 10.82 | -135.73 | 3 |
| 6 | 3369 → 3395 | -19.04 | 21.44 | 10.72 | -51.21 | 26 |
| 7 | 3427 → 3435 | -103.60 | 21.26 | 10.63 | -135.49 | 8 |
| 8 | 3454 → 3467 | -54.96 | 21.04 | 10.52 | -86.51 | 13 |
| 9 | 3471 → 3511 | 47.45 | 20.97 | 10.48 | 16.00 | 40 |
| 10 | 3533 → 3535 | -23.34 | 20.93 | 10.46 | -54.73 | 2 |
| 11 | 3544 → 3580 | 64.45 | 20.91 | 10.45 | 33.09 | 36 |
| 12 | 3584 → 3589 | -49.95 | 20.86 | 10.43 | -81.24 | 5 |
| 13 | 3596 → 3606 | -44.42 | 20.70 | 10.35 | -75.47 | 10 |
Candle: closed / confirmed
Calculation: SMA seed available; recursive EMA can be calculated.
Data: Snapshot fetched 11 Oct 2026, 15:52 UTC · older than one candle period
Last closed candle in the disclosed data: 11 Oct 2026, 14:59 UTC
This is a measurement computed from prices. It does not imply a probability of success.
EMAₜ = α × Closeₜ + (1 − α) × EMAₜ₋₁
0.181818 × 83,054.560000
+ 0.818182 × 82,992.414072
= 83,003.713332
Display is rounded; calculations use unrounded numbers.
Coefficient of the complete SMA seed: 0.00 %. This is a weight, not an error bound. Earlier history still contributes after N candles.
EMA 10: 83,012.9582 · EMA 30: 82,927.7360
first EMA above · No new crossover
Equality at 82,328.041304. The threshold itself is not a strict crossover.
Calculation for exactly one new candle, fixed periods and these EMA starting values. Not a predicted price target; the starting candle must be closed.
A hypothetical spot buy and sell using the same starting capital, cash allocation and cost model as the EMA rule. The separate entry/exit prices and optional loss budget do not change the historical rule. Fractional quantities; no exchange minimum size.
Shared assumptions: 10000 USDC · 100 % cash allocation · 0.10 % fee · 0.05 % Slippage
Only together with the assumed stop price.
0 − 19.97 − 9.985 = -29.9551
Standard EMA with SMA initialization. Parity with a deployed unCoded engine is not established here. CSV contains only the disclosed data. For market data, downloading or sharing a link cannot guarantee later availability of the same snapshot.
Traceable to the formula
N is an integer count of candles, not calendar duration. Alpha = 2/(N+1). The first EMA value is the mean of the first N closes; earlier values are unavailable. Then:
EMAₜ = EMAₜ₋₁ + α × (Closeₜ − EMAₜ₋₁)
For N=1 the EMA equals the close. The seed contributes a coefficient (1−α)ᵏ after k updates. This is a weight, not forecast uncertainty. Zooming does not change the calculation history.
Spot long/cash: a confirmed crossing from below to above creates a buy intent; the reverse creates a sell intent. Execution is no earlier than the next open, including the selected costs. An already-above state does not trigger a catch-up entry.
The evaluation window starts in cash. No leverage, shorts, hidden stop or pyramiding. Each entry allocates a fraction of available cash including its entry fee. Quantities are theoretical; exchange lot sizes are not modeled.
Fees apply per executed side. Slippage worsens each reference open; it includes spread, which is not added twice. Gross P&L uses the same actual quantities and reference opens before slippage. Gross minus fees minus slippage equals net.
A final open position is marked at the last confirmed close. No fictitious sale or exit fee is added. Drawdown is measured at candle closes and misses intrabar troughs.
EMA smooths past prices. A higher value or crossover is not a calibrated probability. The last 30 days end at the snapshot’s last complete UTC hour. 3000 warmup candles only initialize the averages. Variants and benchmarks use the same evaluation window. Repeatedly tuning to the same data can overfit.
The mathematical convention is checked against documented references. Parity with a currently deployed unCoded engine is not claimed. The separate position calculation is hypothetical and does not alter the crossover rule.
Existence and relevant content checked on 11 October 2026. Research, technical references and data documentation serve different purposes; none is used as proof of profitability.
Rob J. Hyndman & George Athanasopoulos · 2021
Recursive exponential smoothing, decaying weights and dependence on the initial level. Does not establish a trading advantage.
TA-Lib contributors; Mario Fortier · 2025
Default compatibility: alpha = 2/(N+1), SMA initialization and recursive update for N ≥ 2. The N=1 identity here is tested separately.
pandas development team
Span-to-alpha conversion and recursive adjust=False convention. The tool declares its SMA seed explicitly.
David H. Bailey, Jonathan M. Borwein, Marcos López de Prado & Qiji Jim Zhu · 2015 · revised manuscript, 27 February
Why selection across many historical trials can produce misleading results. This tool does not estimate PBO or claim out-of-sample profitability.
Binance
Timestamped OHLCV candle schema and intervals. Data origin does not validate a trading rule.
Binance
Public data-api.binance.vision endpoint without account credentials.