quick rant about "constructing a neutral exposure" while I wait for something to load. a qualitative description about step by step optimization for us peons who take seconds to calculate (not microseconds on a custom ASIC like some here)

1/ Have a quant strategy. What's that?
2/ A quant strategy is a thing that assigns an expected value to many assets over a short period of time. It justifies itself primarily out of a process working historically specifically *having been able to predict profits*. A bit on what that means/ why it's important
3/ Basically - if you have the Russell 3k and split it into 50 buckets, a good quant strategy should have bucket 50 work *out of sample* better than 49 work better than 48... 2 better than 1. Having good quartile profitability allows you to have some notion of expected value.
3/ I tend to make expected value a sharpe ratio. These can be multiplied by implied volatility or regime specific realized vol (i.e. a stock going into earnings, or over a weekend etc) to give you a picture of expected return per stock. Sharpe*volatility = expectancy
4/ Put simply - you have a strategy. It has a profitability ratio given a volatility. A stock has a likely volatility. You assume that the profitability ratio will broadly map onto the stock's volatility. You feel comfortable assuming this bc it has been very true in the past
5/ So onto the cooking. the first "Step" is to get everything illiquid into the first basket with capped liquidity assumptions. My first basket is always illiquid garbage. Generally this is very "profitable" on an IRR basis but not on an NPV basis, so get it out of the way
6/ Step 2: identify a hedge for the basket. I only hedge with ETFs (for various reasons, the most significant being they're diversified, the secondary being they tend to really implode vs NAV when SHTF). Identify a hedge w expectancy and realized correlation. Huh?
Basically you're saying [Basket of Garbage] tracks $ARKF with a .63 correlation, and this correlation historically has not dipped below .4. I use "variance adjusted correlation" - so take 1 std less average and realized correl. Re expectancy I map expectancy to ARKF holdings
7/ Step 3: You've got a basket of pebbles + a hedge. Now add rocks. Add 30 stocks, expectancy ranked. Then add a ETF short, with expectancy mapped to its holdings. Now you have a portfolio return that *is additive with your new additions* Add 30 more + ETF short til 80% deployed
8/ Step 4: The last 20% of gross exposure I use to "kill gap risks". First I get the basket's S&P 500 beta to below 0. Next, I have pairwise factor exposures (like MTUM vs IWN). Anything with adjusted corr >.35 gets hedged - once again by mapping expectancy to ETFs
9/ Step 5: You've got your 80% basket and your 20% basket hedges with no expected market or factor exposures. Now you have to blast out orders in a *backtested way*. backtesting execution assumptions is very important!! But Usually ends with some mix of limit/market orders
10/ A note here: chances are you'll be partially filled on a decent % of your book. In practice this means you'll have cancelled limit orders and lopsided market exposures randomly different from your ideal basket. So Step 6: a mini script to "get flat" on beta, factors!
11/ Step 7] Tag the fills with a name so you can attribute your PNL correctly later and measure your realized slippage. If you're running multiple strategies this is important - and worth comparing out of sample PNL to expected as a daily KPI for all strategies
12/ Anyways, I'm sure this is all terribly "wrong" and someone will educate me about the right way to do it. But this is how I approach automatically deploying risk on a quant basis, the goal being to harvest an expectancy with as little noise as possible (aka make $ every week)

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Alex Good

Alex Good Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @goodalexander

13 Oct
Just going to share some interesting quotes from JPM earnings, one of my favorite indicators for earnings season
1/ Double Digit Spend Growth in 3q vs 19. "Combined credit and debit spend was up 24% versus the third quarter of '19. Softened in Aug and reacceled in recent weeks."
2/Seem Convinced Delta is Done. "we see some signs of life and we believe that recovery is strongly underway and it seems hopeful like Delta is really fading so that's going to help."
3/ Credit Quality is Improving a Lot. "But it's also worth noting that net charge-offs of just over $500 million were approximately half of last year's third-quarter number.
Read 14 tweets
13 Oct
Put 3% in Kohl's $KSS @ $45.1. Shorted 3% Consumer Staples $XLP @ 69.99. Overwrote $KSS Jan 21'22 50 Call for $2.48 (5%). Market is over-discounting Kohl's long term story & huge buyback with a short term supply chain narrative.

1/ Debt holders are unconcerned ImageImageImage
2/ A double downgrade out of BoFa on supply chain concerns 'derailing' turnaround plan has caused Kohl's equity to implode versus other retail comps and elevated near term equity volatility. Per the charts above, however, this has not flowed into the company's LT financing costs
3/ Carry. The yield of this trade is extreme. Kohl's has likely 300m of buyback auth into end of the year - an 18% annualized yield on top of the overwriting yield of 22%. When you net out consumer staples capital return of 5-7% annualized leaves you 33% of annual low beta yield
Read 13 tweets
12 Oct
I took a .5% position in Roblox $RBLX Nov 19'21 75 Calls for $3.5 to play a strong Nov 16 investor day. Data tracking in spite of back to school and a metaverse focused catalyst calendar I think can send the stock back to the $85 area

1/ Exhibit 1: Web Tracking data is surging Image
2/ This spike in traffic is surprising and corresponds with good app store performance. In Roblox's September investor day the CFO cautioned investors that back to school might see some decelerating usage. After seeing a brief dip to #50 in iOS we are stabilizing back near #20 Image
3/ Roblox sells Robux tokens to players. Investors appreciate in game currency as an effective revenue model. For example, the crypto game Axie Infinity has 1.5 million daily active players and a $32 billion market cap compared to Roblox 46 million players and $40b market cap ImageImage
Read 17 tweets
11 Oct
I am long BTC @ 57,400. 2.5% of capital with 3 catalysts in mind to move Lightning story forward which builds BTC maxi narrative
Twitter/SQ Earnings: Oct 26, Nov 4
Nov 18: El Salvador Lightning Conf @GaloyMoney

1/ Lightning Capacity will likely be a hockey stick into the events
2/ Lightning is important because
a] it enables global adoption of Bitcoin as currency, powering things like El Salvador's Chivo wallet
b] Layer 2 allows apps - including NFTs to run directly on bitcoin, creating a story for fees can displace the block reward in the long run
3/ ETH story around NFTs, artwork likely is forgetting things like Satoshi.Place pictured below built on top of lightning network. Layer 2s provide theoretically higher TPS and interoperability with Twitter via strike.me . Twitter already verifying NFTs
Read 15 tweets
5 Sep
Some markets related reading I've done past 2 months I found actionable/interesting:
1/ IESE Biz School Paper on equity risk premium in 88 different countries in relation to risk free rate. Nice because it has standardized results going back to 2008. poseidon01.ssrn.com/delivery.php?I… Image
2/ Paper on run up into the close -- hedging intraday demand -- fairly robust discussion of all major asset markets. TLDR it's worth studying how momentum/ mean reversion plays out statistically in stuff you trade going into the close. long backtest
www3.nd.edu/~zda/intramom.… Image
3/ Marc Andreesen on crypto. Think we're seeing some (rumored) monster venture flows into the space next few months which is part of price action and this piece helped catch decent amount of ETH move marginalrevolution.com/marginalrevolu…
Read 9 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!

:(