The Colombian peso is at its strongest since 2018. Almost nobody is asking why that’s not necessarily good news today.
The Colombian peso hit its strongest level since October 2018 this week (3,053.48 pesos per dollar on August 19), with the dollar down 18.5% year to date in 2026. For an importer or a company with dollar debt, the immediate read is relief. But that strength coexists with a fiscal fracture that helps explain it: the magnitude-7.4 earthquake on August 10 — over 289 dead — imposed a $9.5 billion reconstruction cost on a government barely three days into power. Abelardo de la Espriella took office on August 7 promising to bring the deficit down from 6.4% to 4.8% of GDP with $70 trillion pesos in spending cuts, and he’s already conceding “we’re going to take on debt” to fund the emergency. The decision that matters for anyone managing risk in Colombia: a strong peso today is no guarantee of fiscal room tomorrow, with external debt at 52.7% of GDP. The real test for this government wasn’t the inauguration speech — it’s whether the fiscal rule survives its first exogenous shock. If you manage treasury or peso-denominated debt, we already covered why it doesn’t pay to plan on Colombian rates easing soon in this column on the Banrep rate.
AI beat the classic 60/40 portfolio over two decades of simulation — with one key caveat
JPMorgan built AI agents that, in two-decade backtests, beat the traditional 60/40 portfolio by 0.7 percentage points annually — with lower volatility, according to research led by strategist Thomas Salopek. The agents rotate autonomously between stocks and bonds based on the market “regime” they detect, and it’s the bank’s first attempt at building an AI system for regime identification. The firm itself flags the caveat: these are backtests, not a real track record, and its strategists warn against “uncritically accepting overfit in-sample AI answers.” The practical implication: institutional portfolio management is moving toward adaptive systems, but professional judgment — not the backtested result — is still what separates a useful signal from a well-dressed statistical mirage.
The idea that “Wall Street already moved to the blockchain” is making the rounds
The noise: headlines announced “Wall Street moves to the blockchain” when DTCC — custodian of $114 trillion in U.S. securities — kicked off tokenization tests with 40-plus firms (JPMorgan, Goldman Sachs, BlackRock, Citadel Securities) trading tokenized stocks, ETFs, and Treasury bonds.
The signal: DTCC’s own CEO, Frank La Salla, was explicit: this isn’t disruption, it’s freeing up liquidity trapped inside the existing system. The load test used known, low-risk assets (Microsoft, the QQQ ETF, SPY, Treasury bills) on a private, permissioned network (Besu and Canton Network), not an open blockchain — and the full launch doesn’t arrive until October. The tokenization market moves roughly $38 billion today; Citi projects $5.5 trillion by 2030. That gap is the real measure of how early this still is: a well-designed infrastructure test, not the revolution the headline announces.
“Asset tokenization and the use of digital blockchain are a megatrend. What we’re really focused on is the security of the system, its resilience, and freeing up trapped liquidity using this new technology.”
— Frank La Salla, President and CEO of DTCC. While Wall Street celebrates its first tokenized production trades, this quote is a reminder of what’s actually being tested: not the promise, the infrastructure.