🌊 A “stock-to-flow debacle” has taken crypto-twitter by storm.
The meme achieved superior performance in nocoiner conversion…
…and flooded your timeline with 🔢math, 💎minerals and 🔥heated opinions you may not understand.
Time to slash FUD and review some of the #stats:
The meme achieved superior performance in nocoiner conversion…
…and flooded your timeline with 🔢math, 💎minerals and 🔥heated opinions you may not understand.
Time to slash FUD and review some of the #stats:
S2F is a measure of scarcity applied to commodities.
“in how many yrs, at current production levels (flow), will the market add *new assets* equivalent to all current outstanding assets (stock)?”
@TraceMayer, @Saifedean flirted w/ applying it to BTC...
goldstandardinstitute.net/2011/07/16/sto…
“in how many yrs, at current production levels (flow), will the market add *new assets* equivalent to all current outstanding assets (stock)?”
@TraceMayer, @Saifedean flirted w/ applying it to BTC...
goldstandardinstitute.net/2011/07/16/sto…
@TraceMayer @saifedean But it was @100trillionUSD who finally laid out the digital SoV thesis in a language that appealed to money managers worldwide.
Visualising BTC’s S2F X other assets also helped.
In the instantly-turned-classic chart, what pops to the eye is Bitcoin’s fierce path to the #moon.
Visualising BTC’s S2F X other assets also helped.
In the instantly-turned-classic chart, what pops to the eye is Bitcoin’s fierce path to the #moon.
@TraceMayer @saifedean @100trillionUSD S2F is a *metric* that applies to most non-renewable commodities.
The S2F *model* is planB’s ingenious touch.
Regressing the S2F of BTC against Mcap, he found a (well fitting) linear correlation.
It makes forecasts when extrapolated. Today, ~ $11.6k
digitalik.net/btc/
The S2F *model* is planB’s ingenious touch.
Regressing the S2F of BTC against Mcap, he found a (well fitting) linear correlation.
It makes forecasts when extrapolated. Today, ~ $11.6k
digitalik.net/btc/
@TraceMayer @saifedean @100trillionUSD @Ivan_on_tech was among the first to articulate and air critique on the model.
He pointed to the possibility of “overfitting”: we’ve only been through 2 halvings, and the sample of data for Bitcoin price history is still quite limited.
He pointed to the possibility of “overfitting”: we’ve only been through 2 halvings, and the sample of data for Bitcoin price history is still quite limited.
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech @100trillionUSD rebutted quickly. You rarely overfit a linear regression over 100+ datapoints.
Illustrating: a model with *only* data from 2009-2012 would’ve produced good enough predictions till today.
Illustrating: a model with *only* data from 2009-2012 would’ve produced good enough predictions till today.
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech *Overfitting* has a precise definition. Not simply “when a function is too closely fit to a limited set of data” to be reliable.
Green=overfitted, black=generalized 👇
Overfitting is rarely a problem in a linear regression (w/ enough datapoints):
en.wikipedia.org/wiki/Overfitti…
Green=overfitted, black=generalized 👇
Overfitting is rarely a problem in a linear regression (w/ enough datapoints):
en.wikipedia.org/wiki/Overfitti…
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech @Ivan_on_tech might’ve meant the model is “too much well tuned” to BTC.
That’s valid: the *indicator is not new* in physical commodities, but *the model is*.
We’re yet to see the correlation (SF2 <> Mcap) tested for more assets (PlanB did it for 5).
That’s valid: the *indicator is not new* in physical commodities, but *the model is*.
We’re yet to see the correlation (SF2 <> Mcap) tested for more assets (PlanB did it for 5).
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech @100trillionUSD is particularly excited about *cointegration*.
〰️ 2 series being cointegrated (BTC’s S2F and mcap here) = “they move together” (their difference is a stationary series).
If they’re not, one can falsify a regression (deem it spurious).
medium.com/burgercrypto-c…
〰️ 2 series being cointegrated (BTC’s S2F and mcap here) = “they move together” (their difference is a stationary series).
If they’re not, one can falsify a regression (deem it spurious).
medium.com/burgercrypto-c…
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech Cointegration doesn’t PROVE a non-spurious relationship - just fails to disprove it.
In statistics, you don’t prove stuff. You try - and fail - to disprove it, over & over.
No cointegration might have invalidated the model. But the opposite isn't true.
In statistics, you don’t prove stuff. You try - and fail - to disprove it, over & over.
No cointegration might have invalidated the model. But the opposite isn't true.
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech More nuanced critique comes from @IntoTheLoop: he suggests the power law between S2F and Mcap, even if informative, may only reflect there are other common factors co-driving the 2 variables.
How confidently can we really say Scarcity drives MarketCap?
medium.com/coinmonks/a-mo…
How confidently can we really say Scarcity drives MarketCap?
medium.com/coinmonks/a-mo…
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech @intotheloop Partial correlation is a multivariate (not bi-variate like normal correl.) method.
It gauges the lin. relationship b/ween 2 variables *after* removing the influence of a 3rd (and 4th, etc).
S2F contributes little in a classic multivariate regression.
It gauges the lin. relationship b/ween 2 variables *after* removing the influence of a 3rd (and 4th, etc).
S2F contributes little in a classic multivariate regression.
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech @intotheloop @IntoTheLoop is saying Time drives Scarcity and MarketCap to a large extent - hence high correl. between S2F/Mc. There may not be an *extra* driving force by Scarcity on Mcap.
@hcburger1 makes a similar point (see his purely time-based Power Law).
@hcburger1 makes a similar point (see his purely time-based Power Law).
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech @intotheloop @hcburger1 @phraudsta used the Akaike Info. Criterion - an estimator for the quality (signal loss) of models against each other (more adequate than comparing R2s).
AIC for
ln(market price) ~ ln(Stock to Flow)
is lower (better) than for
ln(market price) ~ time
AIC for
ln(market price) ~ ln(Stock to Flow)
is lower (better) than for
ln(market price) ~ time
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech @intotheloop @hcburger1 @phraudsta @IntoTheLoop trolled the crew by suggesting US Industrial Production can also be modelled with BTC’s S2F, and “asking” for cointegration tests to falsify the (obviously spurious) hypothesis.
@phraudsta knocked the joke down with a G-causality test.
@phraudsta knocked the joke down with a G-causality test.
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech @intotheloop @hcburger1 @phraudsta OK!
🧠 So now you’re familiar with the basic concepts around the S2F model.
❌🏃 And also, with how far people’ve come in trying to falsify it.
🗝️ What are the ELI5 TAKEAWAYS?
🧠 So now you’re familiar with the basic concepts around the S2F model.
❌🏃 And also, with how far people’ve come in trying to falsify it.
🗝️ What are the ELI5 TAKEAWAYS?
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech @intotheloop @hcburger1 @phraudsta 1️⃣ Bitcoin hasn’t ever had a pricing model debatable at this level.
Metcalfe Law, MV = PQ, you name it...
👶 S2FM is embryonic, but may welcome many skeptic money managers into Bitcoin.
These can now at least envision an answer to:
what drives BTC value? How to quantify it?
Metcalfe Law, MV = PQ, you name it...
👶 S2FM is embryonic, but may welcome many skeptic money managers into Bitcoin.
These can now at least envision an answer to:
what drives BTC value? How to quantify it?
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech @intotheloop @hcburger1 @phraudsta 2️⃣
All models are wrong. Some are more useful than others.
🎯 Don't expect any S2F price forecast to be exact in a particular date.
💪 Expect the model to be wrong with certain frequency. Also, to improve with new data (this is still out for jury).
All models are wrong. Some are more useful than others.
🎯 Don't expect any S2F price forecast to be exact in a particular date.
💪 Expect the model to be wrong with certain frequency. Also, to improve with new data (this is still out for jury).
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech @intotheloop @hcburger1 @phraudsta 3️⃣
The premise that “BTC is a SoV, hence the model should work”, may be backwards.
Given that we only model things that are a SoV with S2F, BTC may be demonstrating to be a SoV partly *because* S2F model works for it.
The premise that “BTC is a SoV, hence the model should work”, may be backwards.
Given that we only model things that are a SoV with S2F, BTC may be demonstrating to be a SoV partly *because* S2F model works for it.
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech @intotheloop @hcburger1 @phraudsta Insufficient evidence of a strong correlation b/ween S2F and Mcap for Litecoin highlights this point.
LTC may not have the quality of *unforgeable scarcity* (h/t @NickSzabo4), and has not been established as a SoV.
(this hypothesis needs + tests)
medium.com/@phraudsta/lit…
LTC may not have the quality of *unforgeable scarcity* (h/t @NickSzabo4), and has not been established as a SoV.
(this hypothesis needs + tests)
medium.com/@phraudsta/lit…
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech @intotheloop @hcburger1 @phraudsta @NickSzabo4 4️⃣
Some get uncomfortable w/ a model that predicts $50K+ already <1y from now (& much +, further).
To get around that, you can blame the fate of the U$ dollar (hyperinflation)...
...use kWh for measuring purchasing power of BTC (h/t @SGBarbour)...
Some get uncomfortable w/ a model that predicts $50K+ already <1y from now (& much +, further).
To get around that, you can blame the fate of the U$ dollar (hyperinflation)...
...use kWh for measuring purchasing power of BTC (h/t @SGBarbour)...
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech @intotheloop @hcburger1 @phraudsta @NickSzabo4 @SGBarbour …or you can choose to completely disregard a model which eventually predicts bitcoin having infinite USD value (when “flow” ends).
In fact, some of the most sensible critics so far have to do with the flow denominator.
In fact, some of the most sensible critics so far have to do with the flow denominator.
@TraceMayer @saifedean @100trillionUSD @Ivan_On_Tech @intotheloop @hcburger1 @phraudsta @NickSzabo4 @SGBarbour 👋 That was it, folks.
Devour the texts above.
Think independently.
Follow @100trillionUSD, @phraudsta, @hcburger, @IntoTheLoop & @BurgerCryptoAM to keep up with their latest findings. Correct me where I'm wrong!
& thanks to them for putting all out for public scrutiny 🙏
Devour the texts above.
Think independently.
Follow @100trillionUSD, @phraudsta, @hcburger, @IntoTheLoop & @BurgerCryptoAM to keep up with their latest findings. Correct me where I'm wrong!
& thanks to them for putting all out for public scrutiny 🙏
* @InTheLoopBTC is the actual handle you should follow 🔔😔, wrote the wrong one above
* and @hcburger1 not @hcburger 😋
