quantifytools

Session candles & session reversals (+ Backtest)

quantifytools Premium 已更新   
— Overview

Like traditional candles, session based candles are a visualization of open, high, low and close values, but based on session time periods instead of typical timeframes such as daily or weekly. Session candles are formed by fetching price at session start (open), highest price during session (high), lowest price during session (low) and price at session end (close). On top of candles, session based moving average is formed and session reversals detected. Session reversals are also backtested, using win rate and magnitude metrics to better understand what to expect from session reversals and which ones have historically performed the best.

By default, following session time periods are used:

Session #1: London (08:00 - 17:00, UTC)
Session #2: New York (13:00 - 22:00, UTC)
Session #3: Sydney (21:00 - 06:00, UTC)
Session #4: Tokyo (00:00 - 09:00, UTC)

Session time periods can be changed via input menu.

— Reversals

Session reversals are patterns that show a rapid change in direction during session. These formations are more familiarly known as wicks or engulfing candles. Following criteria must be met to qualify as a session reversal:

Wick up:
Lower high, lower low, close >= 65% of session range (0% being the very low, 100% being the very high) and open >= 40% of session range.

Wick down:
Higher high, higher low, close <= 35% of session range and open <= 60% of session range.

Engulfing up:
Higher high, lower low, close >= 65% of session range.

Engulfing down:
Higher high, lower low, close <= 35% of session range.

Session reversals are always based on prior corresponding session, e.g. to qualify as a NY session engulfing up, NY session must have a higher high and lower low relative to prior NY session, not just any session that has taken place in between. Session reversals should be viewed the same way wicks/ engulfing formations are viewed on traditional timeframe based candles. Essentially, wick reversals (light green/red labels) tell you most of the motion during session was reversed. Engulfing reversals (dark green/red labels) on the other hand tell you all of the motion was reversed and new direction set.

— Backtesting

Session reversals are backtested using win rate and magnitude metrics. A session reversal is considered successful when next corresponding session closes higher/lower than session reversal close. Win rate is formed by dividing successful session reversal count with total reversal count, e.g. 5 successful reversals up / 10 reversals up total = 50% win rate. Win rate tells us what are the odds (historically) of session reversal producing a clean supporting move that was persistent enough to close that way too.

When a session reversal is successful, its magnitude is measured using percentage increase/decrease from session reversal close to next corresponding session high/low. If NY session closes higher than prior NY session that was a reversal up, the percentage increase from prior session close (reversal close) to current session high is measured. If NY session closes lower than prior NY session that was a reversal down, the percentage decrease from prior session close to current session low is measured.

Average magnitude is formed by dividing all percentage increases/decreases with total reversal count, e.g. 10 total reversals up with 1% increase each -> 10% net increase from all reversals -> 10% total increase / 10 total reversals up = 1% average magnitude. Magnitude metric supports win rate by indicating the depth of successful session reversal moves.

To better understand the backtesting calculations and more importantly to verify their validity, backtesting visuals for each session can be plotted on the chart:


All backtesting results are shown in the backtesting panel on top right corner, with highest win rates and magnitude metrics for both reversals up and down marked separately. Note that past performance is not a guarantee of future performance and session reversals as they are should not be viewed as a complete strategy for long/short plays. Always make sure reversal count is sufficient to draw reliable conclusions of performance.

— Session moving average

Users can form a session based moving average with their preferred smoothing method ( SMA , EMA , HMA , WMA , RMA) and length, as well as choose which sessions to include in the moving average. For example, a moving average based on New York and Tokyo sessions can be formed, leaving London and Sydney completely out of the calculation.

— Visuals

By default, script hides your candles/bars, although in the case of candles borders will still be visible. Switching to bars/line will make your regular chart visuals 100% hidden. This setting can be turned off via input menu. As some sessions overlap, each session candle can be separately offsetted forward, clearing the overlaps. Users can also choose which session candles to show/hide.

Session periods can be highlighted on the chart as a background color, applicable to only session candles that are activated. By default, session reversals are referred to as L (London), N (New York), S (Sydney) and T (Tokyo) in both reversal labels and backtesting table. By toggling on "Numerize sessions", these will be replaced with 1, 2, 3 and 4. This will be helpful when using a custom session that isn't any of the above.

Visual settings example:


Session candles are plotted in two formats, using boxes and lines as well as plotcandle() function. Session candles constructed using boxes and lines will be clear and much easier on the eyes, but will apply only to first 500 bars due to Tradingview related limitations. Rest of the session candles go back indefinitely, but won't be as clean:


All colors can be customized via input menu.

— Timeframe & session time period considerations

As a rule of thumb, session candles should be used on timeframes at or below 1H, as higher timeframes might not match with session period start/end, leading to incorrect plots. Using 1 hour timeframe will bring optimal results as greatest amount historical data is available without sacrificing accuracy of OHLC values. If you are using a custom session that is not based on hourly period (e.g. 08:00 - 15:00 vs. 08.00 - 15.15) make sure you are using a timeframe that allows correct plots.

Session time periods applied by default are rough estimates and might be out of bounds on some charts, like NYSE listed equities. This is rarely a problem on assets that have extensive trading hours, like futures or cryptocurrency. If a session is out of bounds (asset isn't traded during the set session time period) the script won't plot given session candle and its backtesting metrics will be NA. This can be fixed by changing the session time periods to match with given asset trading hours, although you will have to consider whether or not this defeats the purpose of having candles based on sessions.

— Practical guide

Whether based on traditional timeframes or sessions, reversals should always be considered as only one piece of evidence of price turning. Never react to them without considering other factors that might support the thesis, such as levels and multi-timeframe analysis. In short, same basic charting principles apply with session candles that apply with normal candles. Use discretion.

Example #1: Focusing efforts on session reversals at distinct support/resistance levels

A reversal against a level holds more value than a reversal by itself, as you know it's a placement where liquidity can be expected. A reversal serves as a confirming reaction for this expectation.


Example #2: Focusing efforts on highest performing reversals and avoiding poorly performing ones

As you have data backed evidence of session reversal performance, it makes sense to focus your efforts on the ones that perform best. If some session reversal is clearly performing poorly, you would want to avoid it, since there's nothing backing up its validity.


Example #3: Reversal clusters

Two is better than one, three is better than two and so on. If there are rapid changes in direction within multiple sessions consecutively, there's heavier evidence of a dynamic shift in price. In such case, it makes sense to hold more confidence in price halting/turning.

發布通知:
Update

Added alerts, available for following scenarios:

- Confirmed session candle close
- Confirmed session reversal up
- Confirmed session reversal down

Alerts are available for each session individually and grouped for all sessions, e.g. alert will trigger on session #1 close vs. alert will trigger on any session close. Session reversals can also be set to trigger on either up/down directions individually or any direction using one alert.

🔸 Like what you see?

🍥 Check out @quantifytools for more pragmatic, sleek and free tools.

🍥 New releases will drop regularly, follow us to stay in the loop 🔔
開源腳本

本著真正的TradingView精神,該腳本的作者將其開源發布,以便交易者可以理解和驗證它。為作者喝彩吧!您可以免費使用它,但在出版物中重複使用此代碼受網站規則的約束。 您可以收藏它以在圖表上使用。

免責聲明

這些資訊和出版物並不意味著也不構成TradingView提供或認可的金融、投資、交易或其他類型的意見或建議。請在使用條款閱讀更多資訊。

想在圖表上使用此腳本?