So, like I was saying in the previous post, when the question arises: "so tell me, how can you use this HIVE-BOT Social Anomaly Detector thing in real life?" here's another way to answer that question... with a 211% Return and Sharpe of 1.82 over 5 years. (Will that work?)
Here's your Executive Summary in a nutshell:
We call this one the Dual-Horizon2, which is a market-timing model trading a basket of 50ETF's which combines both momentum and mean-reversion into one strategy depending on the sentiment surrounding the S&P 500 thru $SPY. For more reading, just click the link at the bottom, or visit this blogpost written by Tynan which gets a little deeper into the deets.
So, here's how you do it:
First, get connected to our HIVE-BOT product.
(What's HIVE-BOT you ask? The HIVE-BOT is like an “electronic eye” that monitors online financial conversations from investors and traders on social networks. These networks include private chat rooms, public chat rooms, alternative financial social networks, slack and slack-type chat services, IRC channels and message boards. HIVE-BOT delivers 3 factors, called the Social Anomaly Score (or “SAS”), which work like a forward-looking VIX Index for each given symbol in the S&P 500, in addition to Equities, ETF’s, FX and Futures.)
For this study, we were using HIVE-BOT to keep an eye on how people felt about $SPY using firehose feeds from @Twitter and our buddies over there at @StockTwits
For more information on HIVE-BOT click here
Go over here to @Quantopian to learn more about how to set up the algorithmic trading strategy using @Python, and pay attention to the research notes where it explains how it's set up. If you're a member of the Quantopian community, he even shares with you the CODE YOU NEED TO DO THE ALGO YOURSELF. (If you're not yet a member and are interested in algotrading, you should sign up today.) Because then you just have to connect the sample algo shared in to our trial dataset that's available there too... more on that in their forums.
Run some backtests yourself on their platform using the dataset and the algo, and then slowly let the magnitude of what you're watching before your very eyes set in...
Then you should probably going to want to talk with our sales department before your manager comes in and asks you why you haven't done so already, just sayin'.
See? Easy-peasy lemon-squeezy.
Over & out.