Crypto Radar · Backtest v1.6 · Big movers v1.2 · Relationships v1.0 (suite v2.8) · back to Radar

Has the score actually worked?

Replays the live Crypto Radar score over past Luno candles without peeking ahead, then counts how often a signal reached your rally target before your stop. It measures the past; it cannot promise the future.

Ready. Fees: Luno MY exchange taker is 0.60% under RM5,000 of 30-day volume (a round trip is 1.2%) and 0.50% from RM5,000 to RM200,000 (1.0%); maker orders are 0.35% / 0.25% each way (source: guide.luno.com, checked 2026-10-04). The box starts at 1.0, so set it for your own tier. The live bid-ask gap is added on top.

Macro context (optional): USDT dominance, Bitcoin dominance, dollar, gold

The dollar index is fetched automatically from your Worker (free ECB data). Gold is tried from the Worker first (Binance data; Binance blocks Cloudflare Workers with an HTTP 403 or 451 error, so this often fails) and is then taken from Luno's own PAX Gold market (PAXGMYR, gold priced in ringgit), which only reaches back to when Luno listed it, so the gold rules are a short-sample look, not a full test. The Worker file is unchanged in this release (v2.1): no redeploy is needed. USDT dominance and Bitcoin dominance have no verified free long history, so the page uses two stand-ins built from Luno prices: "Market rising" for falling USDT dominance and "Alts leading" for falling Bitcoin dominance. If you have your own daily CSV files you can load them below instead; they override the automatic data. Load daily CSV files to test whether these regimes help. In TradingView, open the chart on the 1D timeframe, use the layout menu at the top right, then Export chart data (TradingView offers export on paid plans; if yours cannot, leave these empty and the macro rules simply stay hidden). Suggested charts: CRYPTOCAP:USDT.D, CRYPTOCAP:BTC.D, TVC:DXY and a spot gold chart such as TVC:GOLD or OANDA:XAUUSD. Any CSV with a time (or date) column and a close column works. Each series becomes a yes/no regime (below its own 50-day average for falling USDT.D, falling Bitcoin dominance and a weakening dollar; above it for rising gold) and is lagged one day so it cannot look ahead. Use market-close data such as TradingView's: a series that is published late (some government indexes post a day or more after the date) would need a longer lag than the one day built in. The page reports the date range it read, so you can check it covers enough years.

Not financial advice. Hit rate = share of signals where price reached the target before the stop. "Median best / worst point" = the typical highest and lowest price inside the window, a data-based guide for profit-take and stop levels. Overlapping signals are not independent, so ranges are widened to compensate. "Hold" = the average result of simply holding the whole window with no stop or target. The rules are fixed in advance, not tuned; because up to 22 are compared at once, a tick mark needs a stricter one-sided 99.77% range (z = 2.84), not 95%. "Alts leading Bitcoin" = more than half of the loaded Luno alt coins beat Bitcoin's 90-candle return (needs the All setting). Macro rules use the automatic feeds or the CSV files above, lagged one day. The "tight-spread" table repeats the test on coins whose live bid-ask gap is 0.5% or less, the ones you could realistically trade. "(N mo)" = calendar months the rule fired in: many signals in few months is weaker evidence than it looks. "vs random entry (same days)" compares each rule with random entries on the days that rule could actually have fired: coins are scored from their 80th candle, and rules that need 100 or 200 candles only count once a coin has them. Today's bid-ask gap stands in for the past, so thin coins are probably still flattered.

Copy results for Claude

Big-mover study

Looks at the days followed by a big rally (for example a coin closing +25% within 7 days) and asks what simple rules showed the day before, and how many false alarms came with each catch. Includes new coins with only a few months of history. A second table per half checks direction: are the same days also followed by big drops?

Ready. Moves are measured on closing prices from the next open, so a one-trade spike on a thin book does not count.

Copy results for Claude

Relationship study: do alts move with Bitcoin, the Nasdaq or the S&P 500?

Measures how closely the average Luno alt coin's daily move lines up with Bitcoin's and with the two big US stock indexes, on the same day and up to five days either side. It describes the past only: it creates no buy or sell rule and cannot pass or fail the strict test above. Correlation (r) runs from -1 to +1: +1 = they always move the same way, 0 = no link, -1 = they always move opposite ways. Under 0.1 is negligible, 0.1 to 0.3 weak, 0.3 to 0.5 moderate, above 0.5 strong. "Bitcoin first, alts 1 day later" means Bitcoin moved and the alts are read the next day: the only kind of link that could ever give notice in advance. A link counts as clear only if its strict range (stricter than 95%, because 11 rows are judged at once) stays clear of zero over the whole history AND in both halves. Thin coins trade less often, so their prices can lag Bitcoin's by a day through stale prices alone: a next-day link is not proof that Bitcoin leads.

Stock files (optional): where to get them, and how the timing works

Both boxes are optional: with no file the page still tests the alts against Bitcoin. The free FRED website (fred.stlouisfed.org) carries daily closes for both indexes: search for the series NASDAQCOM (Nasdaq Composite, from 1971) and SP500 (S&P 500, only about the last 10 years), then on each series page click the Download button above the chart and pick CSV (FRED's own help page lists Excel, CSV, Image and PowerPoint there). If the menu looks different, any CSV option with a date column and a value column will do. After a run, the results show each file's first and last day: if a file covers fewer years than you expected, set the chart to its longest date range and download again. Any daily CSV with a date column and a close (or value) column also works, for example a TradingView export. These data belong to their owners: use them for your own analysis only. US markets close about 20:00 to 21:00 UTC and a Luno daily candle covers a whole UTC day, so each stock day is matched with the crypto move from the previous stock day's candle close to that day's candle close (a Monday covers the weekend). "Same window" overlaps in time; "next window" starts after the stock close. This assumes Luno's daily candle starts at 00:00 UTC; that is not confirmed from Luno's documents, so the page checks the candles it receives and prints a warning under the results if they start at another time.

Ready. Daily candles only. The first run downloads every coin's history (a minute or two for All); a second run on the same page reuses it.