Squeeze Momentum Indicator

[jaredbroad]

Expected Behavior

Supports "John Carter's TTM Squeeze Momentum Indicator"

Actual Behavior

No implementation of "Squeeze Momentum Indicator"

Potential Solution

Implement "Squeeze Momentum Indicator". Add unit tests

Checklist

• I have completely filled out this template
• I have confirmed that this issue exists on the current master branch
• I have confirmed that this is not a duplicate issue by searching issues

[jdharmon]

Is this John Carter's TTM Squeeze?

[jaredbroad]

Yes sorry, edited for clarity

[steve-goswell]

Is there an ETA on this new indicator?

[jdharmon]

@steve-goswell I implemented this a while ago, but haven't had time to get the tests against external data working.

The exact formula for calculating momentum is not public. I based my implementation on what I could find. I think it is close enough to use as an indicator. If you compare my implementation with the indicator in your trade platform, the momentum histogram values will not match. However, increasing, decreasing, and whether you're above or below 0 should be consistent.

Feel free to take a look at Squeeze.cs and SqueezeMomentum.cs in feature-4166-squeeze-momentum.

[steve-goswell]

many thanks! I'll take a look at the code. are there any plans to add this to the list of indicators?

https://www.quantconnect.com/docs/algorithm-reference/indicators

[AlexCatarino]

Hi @jdharmon, what is the source of the formulae you used and the data for comparison?

[jdharmon]

@AlexCatarino Links to the formulas are in the comments.

/// <remarks>
/// Adapted formulas from
/// https://school.stockcharts.com/doku.php?id=technical_indicators:ttm_squeeze and
/// https://freethinkscript.blogspot.com/2009/05/ttm-like-squeeze-indicator-with.html
/// </remarks>

When I developed the indicators, I used data downloaded from AlphaVantage and cross-checked against thinkorswim. In my testing, my indicators matched the indicator in thinkorswim. I only picked a few places in recent history where the squeeze fired or the histogram crossed 0.

For unit testing, I used test data from one of the existing TestData files; spy_bollinger_bands.txt, I believe. I could never seem to find the time to go line-by-line in the spy_squeeze.csv file and validate that the data is correct. The bug may be in the test data or in my code. I'm not sure.

[LouisSzeto]

spy_smi.csv

[LouisSzeto]

spy_smi.csv
Columns: "SPY high" "SPY low" "SPY close" "SMA" "BB high" "BB low" "KC high" "KC low" "Squeeze on?"

Script:

import talib
import pandas as pd

history = pd.read_csv("https://github.com/QuantConnect/Lean/raw/master/Data/equity/usa/daily/spy.zip",
                       index_col=0, names=["open", "high", "low", "close", "volume"])

high = history.high
low = history.low
close = history.close

high_bb, sma, low_bb = talib.BBANDS(close, 20, 2)
bb = pd.concat([sma, high_bb, low_bb], axis=1).dropna()
bb.columns = ['sma', 'bb high', 'bb low']

tr = talib.TRANGE(high, low, close)
sma_tr = talib.SMA(tr, 20)
upper_kc = sma + sma_tr * 1.5
lower_kc = sma - sma_tr * 1.5
kc = pd.concat([upper_kc, lower_kc], axis=1).dropna()
kc.columns = ['kc high', 'kc low']

smi = pd.concat([bb, kc], axis=1).dropna()
smi['squeeze on'] = (smi['kc high'] > smi['bb high']) & (smi['kc low'] < smi['bb low'])
smi["high"] = high
smi["low"] = low
smi["close"] = close
smi = smi.iloc[:, [6, 7, 8, 0, 1, 2, 3, 4, 5]]
smi.to_csv("spy_smi.csv", header=False)