Pyth Network

Each one is arithmetic over the bars above — what the prices did, restated. None of them is a view about what happens next.

• SMA (20)

  The mean close of the last 20 bars, recomputed each bar. On this chart a bar is one hour, so 20 bars is 20 hours -- a period counts bars, never calendar days.

• EMA (20)

  Weights recent bars more heavily than a simple average of the same length. Seeded from the first 20-bar mean, the textbook convention -- values differ slightly from a platform that seeds from the first bar instead.

• RSI (14)

  Average gain divided by average loss over 14 bars, scaled to 0-100, using Wilder's own smoothing. A flat window with no movement either way reads 50 rather than either extreme.

• MACD (12, 26, 9)

  The gap between a 12-bar and a 26-bar exponential average, with its own 9-bar average beside it and the difference as a histogram. The figures are in the chart's own currency, so the scale moves with the price.

• Bollinger Bands (20, 2)

  A 20-bar average with bands 2 population standard deviations either side of it. The bands widen and narrow with how much the closes in that window varied.

• ADX (14)

  ADX measures how one-directional the last 14 bars were, without regard to which direction. +DI and -DI are the upward and downward components it is built from.

• Stochastic (14, 3, 3)

  Where each close sits within the high-low range of the last 14 bars, as 0-100, smoothed over 3 bars. %D is a further 3-bar average of %K. A flat window reads 50.

• Supertrend (10, 3.0)

  A band 3.0 average true ranges from the bar midpoint, measured over 10 bars. It ratchets one way while the close stays on one side of it and flips to the other side when the close crosses.

• Anchored VWAP

  A cumulative volume-weighted average price, running from the start of the loaded range. On a market that never closes there is no session for a VWAP to reset at, so it runs from an anchor rather than from an opening bell.

• ATR (14)

  The average size of a bar's full range over 14 bars, counting any move from the previous close. In the chart's own currency, so it scales with the price.

• CCI (20)

  How far the typical price sits from its own 20-bar average, divided by the mean absolute deviation of that window and by Lambert's 0.015 constant.

• Williams %R (14)

  Where each close sits within the high-low range of the last 14 bars, on a 0 to -100 axis: 0 at the top of the range, -100 at the bottom. The stochastic's raw line on an inverted scale, read on its own conventional axis.

• Aroon (25)

  How recently the highest high and the lowest low of the last 25 bars occurred, each as 0-100. Period 25 is Tushar Chande's original; TradingView ships 14, so readings here will differ from a TradingView chart left on its default.

• On-Balance Volume

  A running total that adds a bar's volume when its close rose and subtracts it when the close fell. It has no natural zero -- the level depends on how much history is loaded, so it is read as a shape beside the price, not as a number.

• Parabolic SAR (0.02, 0.02, 0.2)

  Wilder's stop-and-reverse level, accelerating toward price from 0.02 in steps of 0.02 up to 0.2. Drawn as separate marks rather than a line because the level jumps when it flips sides.

• Ichimoku Cloud (9, 26, 52)

  Range midpoints over 9, 26 and 52 bars. The cloud is drawn only as far as the last bar: Senkou A and B are normally projected 26 bars ahead, and those bars have not happened yet. Chikou stops 26 bars before the end for the same reason. On a market that never closes, 26 daily bars are 26 calendar days -- closer to Hosoda's original intent than the exchange week he wrote it for.

• Pivot Points (Standard)

  Standard (floor trader) pivots from the PREVIOUS BAR's high, low and close. These were defined from a previous trading session; this market has none, so on this chart they are computed from the previous hour, not the previous day. Fibonacci, Camarilla and Woodie pivots give genuinely different levels and are not implemented.

• Mansfield RS vs Bitcoin (200)

  This coin's close divided by Bitcoin's, as a percentage above (+) or below (-) that ratio's own 200-bar average. Zero-centred, so it is comparable across coins. 200 counts bars of this chart's timeframe; on a daily chart that is 200 calendar days, because this market has no closed days -- the same count spans about 280 calendar days on a stock chart. A descriptive comparison of two price series.