An aging trope was that coding models produce the most intelligent solutions using Python because there was more training data available, but the data shows the opposite...
I don't think this is a result of the quality of the pre-training corpus; all of these evaluations are in completely novel interactive environments. Pre-training datasets are becoming less relevant to how post-trained models think.
Im sorry. $BTC looks like ass fundamentally.
-The @coinbase premium hasnt been positive since May 17th bc @saylor's $STRC ponzi is dead in the water after $BTC selling + $MSTR dilution.
- $STRC is ex dividend today. Today's close at $91.79 is he lowest level since inception.
-Funding rate continues to be positive as people try to "buy the dip"
- Equities continue to cannibalize crypto volume and liquidity. While SPX and NQ are at ATH, $BTC cant make a higher low. I think this bodes poorly for when they ultimately correct.
Everything i see on SpaceX IPO looks like one of those ultra scammy 1% float high FDV coins where the team farms all the airdrop for themselves and then gets all exchanges to list on day one for maximum exit liquidity, and then pay MMs to crime the chart up until the team unlocks
If Strategy sells MSTR to pay the dividend, it's a ponzi.
If Strategy sells Bitcoin to pay the dividend, it's a death spiral.
If Strategy sells STRC to buy Bitcoin, the common stockholders are being "diluted".
If Strategy uses the USD reserve to pay off the debt, it's a "fatal error".
At some point you have to realize the bear thesis has become a Choose Your Own Adventure book for people who hate the ending.
Alright, here’s my $MU thesis:
In 100 years, Micron Technology invents time travel.
And right now, people from the future are coming back in time to buy the stock before it goes completely parabolic.