Sabitlenmiş Tweet
open.substack.com/pub/thequantat…
Hot take: There is a lot of Art that goes into making ML models.
Which is why everyone's player strength model are both correct and incorrect. Just depends how you look at them...
English
Sharps Research
223 posts
Sharps Research
@sharps_research
machine learning purist, focused on interpretability and advanced ml engineering concepts for NBA moneyline https://t.co/ljVS7GpPhA
Montréal, Québec Katılım Mayıs 2024
392 Takip Edilen1.1K Takipçiler
@Devin_J_Monahan @horsewater @paulg So. You say: single family home owners see a house as a basic need and value it differently. Housing is also in the category of "essential", similar to water, food, shelter.
Do a hypothetical. What happens to the price of housing if investors own 10%, 50% 80% of the market?
English
If the goal is to reduce the price of housing, and not to maximize the number of homeowners, this is a bit of a non sequitur.
The value proposition of owning a housing unit is different for a landlord and an occupier. For the landlord the value of the home is not more than the present value of its expected future cash flow. For an owner-occupier, owning the home is also an investment, but more importantly it provides the basic need of housing. If that can be got for cheaper by renting, it's not a bad thing if not everyone owns their own home.
Beyond the calculation of rent vs. mortgage+tax+insurance, owning your residence has upkeep costs in time & labor, and in money for things you can't do yourself. In this light, many people are frankly not cut out to own & manage their own home even if it's a financially viable option vis-à-vis renting. Most college students, for example, should probably be renting. Hell, I know people who own properties and rent a primary residence from someone else because it's convenient.
English
It's so weird how difficult it is for the left to understand that all you have to do to decrease housing costs is to let people build more housing.
M. Nolan Gray 🥑@mnolangray
For 45 years, Berkeley built virtually no new housing. By the mid-2010s, it was the most expensive college town in America. Shortly thereafter, YIMBYs took over and kicked off a building boom. Today, nominal rents are below 2018 rates—remarkable progress on affordability.
English
@horsewater @paulg And.... When they buy them, what happens to the overall prices of houses in the market?
English
@sharps_research @paulg Oh, no, it is that simple. If investors are buying houses in a jurisdiction, that’s because the number of houses being built is insufficient, and their investments would go bad if things changed. They say so themselves in their SEC filings.
English
Housing prices have increased. But the truth is, increasing supply (more homes) doesn't work to lower house prices
When investors compete against families, investors win
You can increase supply, but if the % of homes owned as investments increase, like is has been, you're fucked
Paul Graham@paulg
It's so weird how difficult it is for the left to understand that all you have to do to decrease housing costs is to let people build more housing.
English
The chart crime.
Basketball University@UofBasketball
Playoff Halfcourt Defenses… What stands out?
English
@mwseibel My bad then. I read it wrong. I am a big fan, wasn't trying to be a dick.
English
@sharps_research I don’t think the solution to all problems is to build a startup. That wasn’t the message I was trying to share.
English
Watching victim politics and populism capture both side of the political debate sucks.
English
@mwseibel Mike. To be fair, most startup founders have some sort of privilege in their life, money, friends, network, time, etc.
For lots of people that don't have this, which is a lot... "just overcome and build a company", is kind of tone deaf.
English
Being told you’re a victim your whole life is not great way to motivate a people to overcome challenges and build something better.
English
Home court advantage:
is team specific, and changes with time
The blanket 5% etc. HCA is false.
in reality its a distribution of HCA that averages out around 5%.
Some teams might have a 10% home court advantage. Others closer to 2%
Slicing residuals is pretty powerful
English
Lots of interesting points here - I’ll post a summary tomorrow.
English
I’m a confused Democrat - internet please teach me something. Do we believe the blue cities in America are run well? If so - share some evidence. If not - what are the top 3 reasons why, and what are the top 3 fixes we should be advocating for? Also please share serious policy not campaign talking points if at all possible.
English
@fitzmagic13 Cappers fail cuz they can't beat the market and eventually their subscribers catch on.
English
After working at Whop for almost 2 years and being involved in the gambling space as a capper, 99 percent of the cappers who fail do so because of one thing.
Greed. They assume that the good times will last forever and develop an ego, often surrounding themselves with yes men who just want to latch off of their success. Just because you have money, doesn’t mean you have to be an idiot with it. Surround yourself with the right people.
English
are you counting the oil and gas burned as input
vs.
the actual electricity delivered, as output
again?
you are aren't you
U.S. Department of Energy@ENERGY
False. Just 15% of CA primary energy consumption comes from ‘renewables.’ More than 75% of CA energy consumption is from oil and gas — that’s higher than the national rate. It is true that CA has embraced anti-hydrocarbon policies. The result? CA has the highest electricity rates in the continental U.S., the highest gas prices, and the highest adjusted poverty rate in America. I wouldn’t be bragging if that was my state.
English
@APEX_Hoops @Fiizop My initial feedback.
What makes it the most transparent player model?
English
English
open.substack.com/pub/thequantat…
Hot take: There is a lot of Art that goes into making ML models.
Which is why everyone's player strength model are both correct and incorrect. Just depends how you look at them...
English
➡️ @PlusEVAnalytics breaks down one of the wildest NBA betting edges 🏀
A Same Game Parlay mistake at an offshore sportsbook turned into a massive win, including a Raptors vs Mavericks game that came down to a late, “meaningless” bucket 💰
New episode of Legendary Loopholes on Circles Off, out now 👀
English
@mikegrib8 Could be cool to viz the residuals on all models in a .PNG with a subplot for each model
English
results are in ladies and gents .. this is predicting the following season net rating, with perfect poss counts for everyone. Now, to be clear, the metrics arent same scale so dont read TOO Much.. but still, I have like total 15 features + did this in 1 day, so happy about that.
Mike Gribanov (PI - Prospect Investigator)@mikegrib8
RAPM's Used Deliberately Over Longer Periods. However, Gribanov Otherwise Fully Annihilated R-squared Tracking. Rudolph Gofart or for short GOBERT. Fully mine, no blend, NBA career metric.. its not amazing compared with EPM/DARKO etc but just for fun.. im about to see how it look
English
I think this is the biggest issue I see out there, overfitting, but not in the classic sense. More so having a small out of sample test set, that you can then just easily tweak knobs until you get something good val metric wise for that small sample.
Issue then becomes people try to use said model on true OOS data and it falls apart. In reality the entire goal of a model or metric like this is to perform as well as possible on truly random un seen data.
English
Two models can have the same MAE, R², etc. whatever you judge them on, but be completely different models in practice. Validation metric tells you how well the model performed against a target. It does not tell you what the output is supposed to communicate.
Sharps Research@sharps_research
open.substack.com/pub/thequantat… Hot take: There is a lot of Art that goes into making ML models. Which is why everyone's player strength model are both correct and incorrect. Just depends how you look at them...
English
@AleatoricFC @jomatech I don't think it was ever gone. Some of this is just insanity.
English
@sharps_research @jomatech was so on point! Founder mode is back!
English
@atulit_gaur Pretty colors and a cool viz look amazing if you don't understand the concepts behind dimensionality reduction.
English
@domluszczyszyn Without model transparency this means absolutely nothing.
English
