Spx 1% low volatility range streaks

Spx is on a low volatility streak, taking a look at how long the streaks usually last and how the current streak relates to past instances. Also looking at Spx returns once the spell breaks – as do probably most others, i expected volatility to pick up, that does not seem to be the case. Bill Luby of Vix And More had a recent post supporting the case for low volatility feeding low volatility on a more long term basis. I myself look at a shorter timeframe but its good to keep longer timeframe in mind and not get overly carried away whenever Vix is asleep.

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import datetime as dt

spy = get_pricing(symbols("SPY"), fields=["close_price", "high", "low"],
start_date=pd.Timestamp("2002-01-01"), end_date=dt.date.today())

spy = pd.DataFrame(spy, index=spy.index)
spy.rename(columns={"close_price" : "close"}, inplace=True)
spy.drop(["high", "low"], inplace=True, axis=1)

def rets(values, shift):
ret = (values.shift(-shift) / values) - 1
return ret

spy["pct"] = spy["close"].pct_change()


Last time Spx had a daily move of +/-1% or greater was on 7th of december

spy[abs(spy["pct"]) >= 0.01].iloc[-1]


Lets look at the previous low volatility streaks. Setting up the streak instances count and index

spy_reset = spy.copy().reset_index()
instances = []

start_index = [0]
end_index = [0]
for index, row in spy_reset.iterrows():
abs_ret = abs(row["pct"])

if abs_ret > 0.01:

if abs_ret <= 0.01:

if abs_ret <= 0.01 and abs(spy_reset["pct"]).iloc[index-1] > 0.01:

instances.append([start_index[-1], row["pct"]])

instances_df = pd.DataFrame(instances)
instances_df.columns = ["inst", "pct"]
instances_df = instances_df[instances_df["inst"] != 0]

Looking at all the streak instances, where Spx daily pct moves have been below 1%, this includes moves both up and down. The current streak is nearing previous record (data starts from 2002). In absolute return terms, its rather close to the previous record and the wall of worry seems to have a consistent return angle across all long low volatility streaks

current_run = instances_df.groupby(["inst"]).get_group(3760)
current_run.loc[-1] = [0, 0]
current_run.index = current_run.index + 1
current_run = current_run.sort()

for index, group in instances_df.groupby(["inst"]):
if len(group) > 10:
group.loc[-1] = [0, 0]
group.index = group.index +1
group = group.sort()
plt.plot(group["pct"].cumsum(), color="#555555", alpha=0.42, linewidth=1, label="_nolegend_")

plt.plot(current_run["pct"].cumsum(), color="crimson", label="Current streak (Start, Dec 12 2016)")
plt.title("Spy low volatility streaks (+/-1%) of longer than 10 days", fontsize=11)
plt.xlabel("# Of trading days in low vol streak")
plt.ylabel("Streak return")
plt.legend(loc="center right")


Now that the streaks are defined, one can look at the corresponding Spx 3 month forward returns when the spell breaks and Spx gets a move of more than 1% up or down. I included only performance fo streaks lasting longer than 21 days. I was expecting to see volatility pick up

returns_df = spy.copy().reset_index()
returns_df["streak"] = np.array(instances)[:, 0]

returns_n = pd.DataFrame(index=np.arange(0, 100))

for index_g, group in returns_df.groupby(["streak"]):
if len(group) > 21:
index_g = int(group.index[-1])
out = returns_df.iloc[index_g:index_g+64]
out = out.reset_index(drop=True)
returns_n[index_g] = out["pct"].cumsum()

returns_n.loc[-1, returns_n.columns.values] = 0
returns_n.reset_index(drop=True, inplace=True)
returns_n.dropna(how="all", inplace=True)

plt.plot(returns_n, color="#555555", alpha=0.42, linewidth=1, label="_nolegend_")
plt.plot(returns_n.mean(axis=1), color="crimson", label="Mean rets")
plt.title("Spy returns after a break of low volatility streak (+/-1%) of longer than 21 days", fontsize=11)
plt.yticks(np.arange(-0.15, 0.10, 0.02))
plt.xlim(0, 63)
plt.xlabel("# Of trading days after low vol streak ends")
plt.ylabel("Post streak return")
plt.legend(loc="upper left")


Thanks your time and feel free to leave a comment


4 thoughts on “Spx 1% low volatility range streaks

  1. Quantocracy's Daily Wrap for 02/21/2017 | Quantocracy

  2. Hi,

    Thanks for your research and this post was great. I’m learning python/pandas and i was wondering if you could explain the lines 15 & 17?

    “if abs_ret 0.01:

    instances.append([start_index[-1], row[“pct”]])”

    Are you saying: “if the abs value of the return is 1% append the start index list with the index of the dataframe? ”

    I’ve run the code in jupyter and i’m thinking that it has to do resetting the streak count, but i would love to hear your rationale!


  3. Hi Darrin and sorry for the late reply
    Im very sure that there is an easier way of counting the steaks. However, you are correct, the point there was to start a new count if today’s return is below 0.01 and yesterdays return was above 0.01

    I would recomment the excellent DataCamp.com courses for learning more python, statistics etc. Drop me an email if you sign up since i will get some sort of discount for my subscription in that case :)


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s