marketrender/crypto-trend-flips

MarketRender crypto trend flips

MarketRender (marketrender.com) is a free, read-only trend map for crypto. Using fixed, published rules it shows whether each coin it follows, and the market as a whole, is bullish or bearish on 4-hour, daily, weekly and monthly candles, in US dollars and in Bitcoin. It keeps a public record of every flip and of what happened afterwards. This dataset is that record, as it stood on 2026-10-08.

It is a record of what happened, not advice: nothing in it says what anyone should do. The site's disclaimer is at https://marketrender.com/legal/disclaimer.

What is in it

• flips.csv.gz (flips.csv, gzip-compressed): 211,091 flips of 268 coins, found from 2017-07-17 to 2026-10-08. One row per flip, with what the price did 7, 30 and 90 days later.
• crypto_trend_score.csv: 3,498 rows of the Crypto Trend Score, MarketRender's whole-market trend score, on daily, weekly and monthly candles, 2018-07-24 to 2026-10-07.

How a flip is defined

A flip is the candle on which a coin's trend line changes sides. The line is a Supertrend with an ATR length of 10 and a multiplier of 3 for Bitcoin and 2 for every other coin, worked out on closed candles only. When a candle closes above the line the state is bullish; when it closes below, bearish. The first state a coin settles into is a starting state, not a flip. A flip is known only when its candle has closed, and that moment is known_at.

The rules are published at https://marketrender.com/about and the read-only API that serves the same data is documented at https://marketrender.com/api-docs. A rule change always gets a new version and a full recompute, and old rows stay, so rule_set tells which version found each flip. US-dollar rows come from flip-line-v1.0; Bitcoin-quote rows from flip-line-v1.1 (the coin's own market against Bitcoin where one exists, otherwise the coin's dollar close divided by Bitcoin's).

Dates and times

All times and dates are UTC. A candle is labelled by the last day it covers, the way the site prints it: Sunday for a weekly candle, the last day of the month for a monthly one. Weekly candles open on Monday 00:00 UTC. Only closed candles are used; weekly and monthly candles are built from the daily ones.

flips.csv columns

Column	Type	What it is
flip_id	integer	Running number of the flip in the MarketRender ledger. It never changes.
coin_id	string	Identifier of the coin, a lower-case slug such as bitcoin or ethereum.
symbol	string	The coin's ticker in capitals, such as BTC.
quote	string	What the price is measured in: USD (US dollars; on Binance, MEXC and Gate that means USDT) or BTC (Bitcoin).
timeframe	string	The candle size the flip was found on: 4h, 1d (daily), 1w (weekly) or 1M (monthly).
candle_last_day	string	The date of the last day of the flip candle, in UTC: the day itself for a daily candle, the Sunday for a weekly one, the last day of the month for a monthly one. For 4-hour candles it is the full UTC time the candle closed (YYYY-MM-DDTHH:MM:SSZ).
known_at	string	The UTC moment the flip candle closed, which is when the flip became known (YYYY-MM-DDTHH:MM:SSZ).
flipped_to	string	The state the trend changed to: bullish (price closed above the line) or bearish (below it).
close	number	The close of the flip candle, in the quote's unit (dollars, or Bitcoin).
rule_set	string	The published rule set that found the flip: flip-line-v1.0 (US dollars) or flip-line-v1.1 (Bitcoin).
change_7_days_pct	number	Price change in percent from the flip candle's close to the daily close 7 days later. Empty until 7 days have passed.
change_30_days_pct	number	The same, 30 days later. Empty until 30 days have passed.
change_90_days_pct	number	The same, 90 days later. Empty until 90 days have passed.
vs_bitcoin_7_days_pct	number	The coin's 7-day change measured against Bitcoin's change over the same days, in percent (positive: the coin did better than Bitcoin). Empty for Bitcoin itself, for rows in the Bitcoin quote, and when not yet judged.
vs_bitcoin_30_days_pct	number	The same, 30 days.
vs_bitcoin_90_days_pct	number	The same, 90 days.
reversed_within_7_days	boolean	true if the same coin flipped again on the same candle size, quote and rule set within 7 days of this flip, false if not. Empty until 7 days have passed.
reversed_within_30_days	boolean	The same, 30 days.
reversed_within_90_days	boolean	The same, 90 days.
scored_now	boolean	true if MarketRender still follows the coin on the generation date, false if it has since been excluded, lost its market or dropped out of the list. False rows are kept on purpose.

What the change columns measure

The change columns are plain price changes in percent, measured from the close of the flip candle to the daily close 7, 30 and 90 calendar days after known_at. They are not adjusted for the direction of the flip: after a bearish flip, a falling price shows as a negative number. A cell is empty when the period has not finished yet or the price needed was not available; nothing is filled in.

crypto_trend_score.csv columns

The Crypto Trend Score blends three parts: Bitcoin's trend, the trend of the other large coins, and breadth (how many of the 200 largest followed coins are in an uptrend). Each part runs from -100 to 100.

Column	Type	What it is
timeframe	string	The candle size: 1d (daily), 1w (weekly) or 1M (monthly).
candle_last_day	string	The date of the last day of the candle, in UTC (Sunday for a weekly candle, the last day of the month for a monthly one).
score	integer	The Crypto Trend Score for that candle: whole-market trend strength from -100 (strongly bearish) to 100 (strongly bullish).
band	string	The score in words: Strong bull (60 and above), Bull (20 to 59), Neutral (-19 to 19), Bear (-59 to -20), Strong bear (-60 and below).
trend	string	bullish or bearish: the direction the score has settled in. It changes only after the score has passed about 35 points on the other side and held there, so it can differ from the sign of the score. Empty while there is not enough history.
bitcoin_part	number	The part of the score from Bitcoin's own trend, -100 to 100 (about 41% of the score).
altcoins_part	number	The part from the trend of the other large coins, -100 to 100 (about 24%).
breadth_part	number	The part from how many of the 200 largest followed coins are in an uptrend, -100 to 100 (about 35%).

Things to know before using it

• Survivorship. Coins that were later excluded or lost their market stay in the file with scored_now set to false, but coins that were never followed are not in it at all, and the followed list itself is today's list of large coins, so older years describe coins that are big now.
• Flips are not independent. Many coins flip on the same days, and a flip is often followed by another one soon after (reversed_within_* shows how often). Counting rows as independent observations overstates the evidence.
• Bitcoin-quote rows have no vs_bitcoin_* figure, since they are already measured in Bitcoin.
• History depth differs. Each coin is read from one exchange, so how far back its 4-hour candles go depends on that exchange and on when the coin was listed. Dollar prices on Binance, MEXC and Gate are USDT prices.
• The score's early years rest on few coins. Before 2021 the breadth part counts between 10 and 80 coins, and the list of the 200 largest followed coins changes over time.
• Provisional closes. Before 5 October 2026 a few Coinbase candles were read just before the exchange had settled them, so a handful of earlier flips rest on a provisional close.
• One snapshot. The files show the ledger as it was when they were generated. Later versions may add rows at the end and fill in outcomes that were empty.

Where the numbers come from

The values are worked out by MarketRender from exchange candles (Binance, Coinbase, Bitstamp, MEXC and Gate) under its published rules. No CoinGecko data (names, market caps, ranks, logos, supplies) is included. The coin_id slug is only an identifier, and which coins were followed was decided from CoinGecko's list of the top 500 coins at the time.

Licence

Licensed under CC BY-NC 4.0: free to share and adapt with credit, for non-commercial use. Credit: "MarketRender (marketrender.com)".

How to cite

MarketRender (marketrender.com). MarketRender crypto trend flips, version 2026-10-08. https://marketrender.com

@misc{marketrender_trend_flips,
  author = {{MarketRender}},
  title = {MarketRender crypto trend flips},
  year = {2026},
  version = {2026-10-08},
  url = {https://marketrender.com}
}

Contact

Questions, corrections and requests for other uses: marketrender@gmail.com. MarketRender is run by an individual based in Lithuania; name and address are given on request by email.