13 · Daily · A wing of the CANDLEWORK network
The Daily Chart Puzzle: what happens next?
One practice chart a day, cut off part-way. You guess whether the price will be higher or lower fifteen candles later, then the rest of the chart is revealed. The charts are simulated, not real market data, and they are built so that nobody can call the direction. That is the lesson: on random data about 50% is the honest score, and convincing patterns turn up anyway.
SIMPractice charts — simulated, not real market data · education only, not advice
Higher or lower in fifteen candles?
Everyone gets the same chart today, made fresh from the date. Read it however you like — trend, levels, candle shapes — then make the call. Your result, score and streak stay in your own browser.
This is a sample practice chart, shown in full because the puzzle needs JavaScript. With JavaScript on, the candles after the dashed line stay hidden until you guess. In this sample the chart had been climbing — and then finished lower.
Fifteen candles from now, will the close be higher or lower than the last close you can see?
Practice charts — simulated, not real market data. Every candle is generated in your browser from a random model; none shows a real instrument, price or period. Education only: nothing here is financial advice or a signal to buy or sell anything. Real trading carries substantial risk of loss.
Your score against a coin
On charts built to be unpredictable, a coin flip scores about 50% in the long run, and so does everyone else. The useful question is not “am I above 50%?” but “am I outside what luck alone would produce?” — and for the first few dozen puzzles, almost nobody is.
Accuracy vs. pure luck
When you felt sure
| You said | Played | Right |
|---|
On random charts every row should drift towards 50%. If your “sure” row is no better than your “coin flip” row, that is the feeling of confidence, measured.
Last 14 days
Your streak counts days you played, not days you were right — showing up is the part you control.
Your browser is not letting this page save, so your score and streak will reset when you leave. The puzzle still works.
Scores and streaks appear here once you have played. They need JavaScript and are saved only in your own browser.
What a chart can and can’t tell you
On these practice charts, the direction of the next move cannot be read from the chart at all, but the size of the moves partly can. Patterns still appear, because randomness draws them. Real markets are not this simple model, and the research on whether chart reading helps there is mixed — so treat every pattern as a description of the past, not a forecast.
Direction
A coin flip, by construction
Every step in these charts is equally likely to go up or down, whatever came before. We tested 100,000 of them: betting that the last 20 candles’ direction would continue was right 49.7% of the time, and betting that a breakout to a new high or low would continue, 50.2%.
Volatility
The width of the swings carries over
The model lets calm and wild stretches cluster. In the same test, after ten unusually wide candles the next fifteen were wider than the chart’s average about three times in four, and calm stretches stayed calm about as often.
Big moves tending to follow big moves was described in real price data by Mandelbrot (1963).
Patterns
Randomness draws convincing shapes
Trends, ranges and “breakouts” appear on every one of these charts, though nothing put them there. Roberts (1959) showed that price series built from random numbers can look like real stock charts, complete with familiar patterns.
Real markets
The evidence is mixed
A survey of the research by Park & Irwin (2007) found many studies reporting profits from technical rules, but most of them open to problems such as testing many rules and keeping the winners, choosing rules after the fact, and underestimating risk and costs. The Contested Room goes through popular methods one by one.
Risk and reward
Plan around the spread, not the call
After each puzzle you can see 200 other futures from the same moment. Their spread — how far the price could plausibly go either way — is the part you can estimate, and it is what position size and stops are for. Being right on direction is worth little if the size of being wrong was never planned.
How the practice charts are made
Each chart is drawn in your browser by a small random model. No market data is downloaded, stored or shown — there is none to show.
- One seed per day. Today’s local date is turned into a number that seeds the random generator, so everyone sees the same chart on the same date and a new one arrives at your local midnight.
- A random walk with moods. Each candle’s move is a random draw with no upward or downward lean. How big the draws are drifts over time: calm stretches cluster, wild stretches cluster, the mood shifts now and then, and an occasional outsized move is thrown in.
- Candles from a path. Each candle is built from a short random path inside its period, which gives it its open, high, low and close.
- Seventy-five candles, fifteen hidden. You see sixty. The other fifteen are already decided before you guess, so your guess cannot change them.
- Two hundred other futures. After you guess, the same model runs from the same moment two hundred more times, to show the range of what could have happened.
Practice charts are not real past charts, are not copies of any real instrument, and do not predict anything. Read the site’s editorial standards for how every simulation here is labelled.
Common questions
Are these real market charts?
No. Every chart is a practice chart, simulated in your browser from a random model seeded by the date. No real market data is used, and none of the charts shows a real instrument, a real price or a real past period.
Why is about 50% the expected score?
The practice charts are built so that every step is equally likely to go up or down, whatever came before. On data like that, no reading of the chart can beat a coin flip in the long run. In a test of 100,000 of these charts, betting on the last 20 candles’ direction was right 49.7% of the time.
Can I get better at the puzzle with practice?
Not at calling the direction, by design. What practice can build is a feel for how often confident-looking patterns fail, how wide the spread of outcomes is, and how streaks of luck feel from the inside.
Does a good score mean I could trade well?
No. A score here measures luck on simulated charts. It says nothing about trading ability, and nothing on this page is a signal or advice to buy or sell anything.
What is saved, and where?
Your daily guesses, how sure you said you were, and your streak are saved in your own browser’s local storage. Nothing is sent to a server. Clearing this site’s data in your browser erases it.
Where the claims come from
The 100,000-chart figures come from running this page’s own model; they describe these practice charts only. The claims about real markets come from:
The Journal of Finance 14(1), 1–10. Random-number price series that resemble real stock charts, patterns included.
The Journal of Business 36(4), 394–419. Early description of large price changes tending to be followed by large changes.
Journal of Economic Surveys 21(4), 786–826. Reviews the evidence and the problems that make it hard to read.