You know, trying to keep pace with economic news these days, it feels uh almost impossible sometimes, doesn't it?

It really does.

And the traditional forecasts, well, they often seem to fall short. Either they arrive way too late or they just lack the the kind of depth you need to make actual decisions.

Yeah. Like you're navigating blindfolded sometimes.

Exactly. Like trying to read a blurry map in a fog. But what if you could actually, you know, see ahead a bit clearer?

That's the million-dollar question, isn't it? And it's precisely why we're doing this. deep dive. Today we're going to unpack this really uh groundbreaking collaboration. It's between Else and Exponential Technologies or Xtech.

Okay.

And it's basically transforming how we predict major macroeconomic announcements. Our whole mission here is to show you how this partnership gives a well a pretty unique advantage insights much earlier and uh with surprisingly good accuracy.

That sounds like a gamecher and we've gathered quite a bit of material for this, haven't we?

We have lots to go through. through.

Yeah. Detailed product info, profiles on the experts involved, uh hard performance numbers, all aimed at giving you a really clear picture of this this new approach. The goal is you walk away with foresight, not just looking back in hindsight.

Let's get into it.

Okay. So, this partnership, Else and Ext.

Well, what's really interesting is that these LEAG global macro forecasts, they aren't just an internal LSE thing. They're built with Xtech. People are calling it a dream team setup.

A dream team, huh? Okay. So, what is each side bring to the table,

right? So, LEG, they bring their their worldclass historical point in time economic data. That's the Pike He data we'll probably talk more about and also their huge database of consensus economic estimates.

So, the deep established data foundation, decades of meticulous records.

Yes, exactly. That's the bedrock. Then, Xtec comes in with uh advanced analytics, cutting edge data tech, and crucially about 25 years of real world institutional investing experience.

Okay, experience matters there. Absolutely. And the person leading Xtech, Morgan Slade, the CEO, he's got like 25 years under his belt as a senior researcher, trader, portfolio manager, places like Citadel, Meil Lynch.

Wow. Serious credentials.

Yeah. Plus engineering degrees from MIT. He was apparently one of the pioneers in using AI to pull actual investment signals out of data.

So, it's not just data meeting data. It's deep data meeting serious brain power and uh cutting edge tech.

That's a great way to put it. And this new forecasting approach, it uses leading indicators, the LEG P data, their own proprietary alternative data and AI. Sounds like a completely different beast from the old models.

It really is. And that's a key distinction they make. These predictions are built uh bottom up.

Bottom up, meaning

meaning based on independent sources, not just refining the existing consensus. This makes their whole approach, well, the word they use is orthogonal to the traditional broker forecast.

Orthogonal. Okay. I think I get that. It means they're tackling it from a completely different independent angle, right? And not just iterating on what everyone else is already doing.

Precisely. They're sort of rethinking the process based on different often unique inputs. It's not just a slight tweak.

That bottomup idea really resonates. How do they keep it sharp though? Make sure it keeps improving. Is there a specific like learning method?

Yeah, absolutely. They use something called teacher forcing or uh sometimes it's called one step ahead forecasting. Key forcing sounds intriguing.

It's actually pretty straightforward in concept anyway. It just means the model is constantly corrected using the latest actual data that comes in. It doesn't just rely on its own previous predictions which might drift off course.

Ah, okay. So, it's always being pulled back to reality essentially learning from the real numbers as soon as they're available.

Exactly. Like a student constantly checking their homework against the answer key, not just guessing based on their last guess. It creates this really powerful continuous learning loop.

That loop sounds key. keeps it agile. Okay. So this innovative approach, what exactly are they predicting with it in this first release? Which indicators?

So right now the focus is on some major US indicators. We're talking the US consumer price index or CPI,

big one,

huge uh US retail sales, also very important.

And then two key measures of how consumers are feeling,

the Conference Board Consumer Confidence Index and the University of Michigan consumer sentiment index.

Got it. key measures of inflation, spending, and sentiment. And I heard something about the CPI forecast being well unique in how it's released.

Yes, that's right. It's not just one single prediction drop. It's multi-stage, which is pretty interesting.

Okay, tell me more.

So, the first CPI forecast, it comes out on the third Monday of the current month.

The current month.

Yeah. Which means it lands more than 3 weeks before the official CPI number is actually released by the government.

Wow. 3 weeks plus. That's That's way earlier than the usual consensus. forecast from brokers, isn't it?

Way earlier. Brokers typically haven't even put out their main predictions yet. This first forecast uses the historical data, recent consumer surveys that have just come out, and that alternative data we mentioned.

Okay, so that's stage one, super early read. What's next?

Then you get a second forecast. This one is released on the third trading day of the following month. So now you're about maybe a week or so ahead of the official release.

And presumably it uses more data that's become available.

Exactly. It incorporates newer information, including things like uh surrogate data from government agencies that might hint at the final numbers. It builds on that initial forecast,

right? Getting sharper as the date gets closer. Is there a third one?

There is. A third forecast drops just two trading days before the official CPI release. This one folds in even more lastminute information.

So, it's like a progressive zoomin. You get an early signal, then a refined one, then a final check just before the real number hits.

That's a good way to think about it. Progressive refinement.

And is it just the main headline CPI number or do they get more specific?

Oh, they get much more specific. That's another valuable piece. You get forecast for individual CPI categories, too. Things like uh gasoline, shelter, food, transportation services, medical.

Gasoline must be a big one. It's so volatile. Drives so much of the headline number sometimes.

Absolutely. And they specifically highlight gasoline as an example where their model has shown, you know, remarkable precision, which is crucial given how much that category can swing markets.

That level of detail is impressive and this is all built on, as you said, deep historical data.

Very deep. So, the forecasts themselves, the track record for CPI starts back in November 2017. For retail sales and the two confidence indexes, it's January 2019.

Okay? Several years of forecast history to look at,

right? But here's the really amazing part. The underlying USCPI data within Else's point in time database, it goes back to 1913.

1913. Wow. Over over a century of foundational data.

Yeah, talk about historical depth. That's serious context.

Okay, so that depth combined with the AI and the early signals, it leads to the core advantage, right? Earlier and more accurate insights.

That's the claim and the numbers seem to back it up. Let's look at some of the performance metrics they've shared focusing on CPI and that conference board index.

Okay, hit me with the stats.

All right, for the headline CPI, month overmonth percentage change, that first forecast, the really early one, it shows an 82% correlation with the actual number. 82% that's strong especially weeks out

it is and 75% directional accuracy so getting the up or down move right 3/arters of the time plus 94% sign accuracy

okay and the second forecast the one about a week out does it improve

it does slightly correlation bumps up to 84% directional accuracy hits 80% and sign accuracy stays at that high 94%.

80% directional accuracy a week out is pretty compelling and what about that volatile gasoline category you mentioned

uh yes gasoline is where it gets uh frankly stunning. Despite the volatility, they report a 96% correlation for the month-over-month change.

96.

Yep. An 87% directional accuracy, 97% sign accuracy. Getting gasoline right that often, that early, is a huge deal for anyone trading energy or inflation expectations.

No kidding. That's incredible precision on a noisy number. What about the Conference Board consumer confidence?

For the Conference Board index level prediction, they show an 89% correlation and 71% directional accuracy. Still very strong. on.

Yeah, definitely. It does make you wonder uh how the traditional forecasters are feeling about this kind of competition.

Well, it certainly seems to raise the bar, doesn't it? Both for accuracy and for timeliness. And that timeliness, the lead time is really significant.

How much earlier are we talking on average?

So, looking at the period from November 2017 through April 2025, the LCG's tech forecasts were available on average 12 days before the consensus forecasts from economists typically solidified

12 days almost two full weeks. That's a massive head start in financial markets.

It's a significant amount of time to analyze, strategize, potentially position

and it's not just earlier. You're saying it's also more accurate. Is there a specific metric for that comparison?

Yes, they use a hit rate metric.

Basically, how often the forecast falls within a tight band around the actual release, they use plus or minus three decimal places for CPI.

Okay. A measure of precision.

Exactly. And the LEG XLA CPI forecasts achieve received a 36% hit rate by that definition, an average of 24 days before the official release.

36% hit rate nearly a month early. How does that compare?

Well, the implication is that it's significantly better than what professional economists achieve, especially considering the lead time. It really highlights that dual advantage

earlier and more accurate.

That really drives the point home. Now, you mentioned point in time data earlier, the PIT data from LEG. Can you unpack why that's so important in this context?

Absolutely. This is crucial. LG's PIT data isn't just a static archive of old numbers. It's a dynamic record. It means you could pull up an economic indicator like CPI or GDP or whatever and see exactly what number was reported on the initial release date at that specific moment in time.

Uh not the revised number that might come out weeks or months later.

Precisely. It captures the first print and then all the subsequent revisions timestamped. Think about testing a trading strategy based on economic surprises. If you use today's revised historical data, You're testing against numbers that didn't actually exist when the market was reacting,

right? You'd have artificial foresight built into your back test. Revision bias.

Exactly. PIT data eliminates that revision bias.

It lets you see history as it actually unfolded based on the information available at that point in time. It's fundamental for building realistic models and truly understanding how markets reacted.

That makes so much sense. It ensured you're learning from the real information flow, not a cleaned up version after the fact.

It's a foundational difference for serious analysis.

Okay. So, bringing this all together, what does this LE Axe Tech forecasting power mean for you, the listener, whether you're an active investor or an analyst, maybe just someone who really wants to understand economic trends?

Well, fundamentally, it offers what you could call a transformativeformational advantage. It's about getting that foresight, seeing around the corner in a way that frankly hasn't been widely possible before.

And how does that translate into practical action? What can people actually do with this earlier, more accurate insight.

Well, there are tons of applications. The most obvious is probably pre-positioning portfolios.

So, adjusting your investments before a big number hits and the market reacts.

Exactly. Getting ahead of the price movement. You can also integrate these macro forecasts directly into trading strategies, maybe directional bets or relative value trades between assets.

What about finding situations where the market might be wrong?

That's another key use. Arbitraging mispricings. If the traditional consensus estimate looks off compared to what these forecasts suggest, there might be an opportunity before the official release corrects things.

Makes sense. And risk management, too, I imagine.

Definitely. You can adjust your portfolio's risk exposure based on anticipated shifts in, say, inflation or economic growth or consumer spending trends revealed by these forecasts.

You could probably use it for broader strategy, too, like asset allocation.

Absolutely.

Yeah.

Leverage these macroeconomic trend insights across different asset classes,

equities, bonds, funds, commodities, currencies.

Yeah,

it helps optimize that big picture allocation.

Better market timing as well, knowing when to get in or out.

Yes, integrating these predictive signals can definitely help refine entry and exit points for trades or investments

and just for pure research, building better models.

For sure, it fuels fundamental research in strategy development with genuinely forward-looking macro modeling, not just extrapolating the past.

It also sounds like it prepares you for surprises.

That's a great point. Being ready for those moment moments when the actual number differs significantly from the consensus, the shocks are disappointments. You might be better positioned to provide liquidity if it's a temporary shock or maybe capitalize if it's a catalyst that breaks a technical pattern.

So really, the core benefit boils down to that phrase we used earlier, gaining the edge of foresight, not just hindsight.

That sums it up perfectly. Anticipation versus reaction.

Okay, so let's quickly recap this LEG and Xtech partnership. It's delivering macroeconomic predictions particular particularly for key US indicators like CPI that are demonstrabably earlier and more accurate than traditional methods.

Right? Leveraging that deep LEG point in time data, proprietary alternative data, and XEX AI and market expertise.

It provides a truly unique source of market intelligence, a realformational edge

indeed. And maybe a final thought to leave folks with,

consider how this combination, the deep history, the alternative data, the AI,

isn't just about getting a better forecast number.

Okay? Think about how it fundamentally changes the speed and the very nature of decision-m for individuals, for institutions. What does it mean when you can glimpse the likely future of key economic indicators with this kind of clarity and this much lead time?

Yeah, that's a big question. What new strategies become possible? How does market behavior itself change when foresight, not just hindsight, starts to become more accessible?

Exactly. It's not just a better tool. It potentially changes the entire game. Something to definitely mull over.