With Almanis, experts’ forecasts can beat group consensus and surveys.
Almanis answers the question …
“Who do I listen to, and when?”
Almanis finds the "Wisdom in crowds".
On Almanis this is the first humans with important new information about future events.
In nature this is the first zebra that has important new information for the group.
Group consensus and surveys find the "Wisdom of crowds".
In nature this is the stampede that follows the first zebra.
For humans, it is the prevailing consensus.
Find out about the Almanis developer, Dysrupt Labs
Dysrupt Labs is a private company driving advanced forecasting methods and is an input to over GBP 5 Trillion in assets under management.
Established in 2008 by two Columbia alumni who as experienced bankers set out to improve forecasting of risk and geopolitics. They used basic and applied research, on their Almanis forecasting platform, in field experiments lasting from 8 hours to 8 years for G7/G20 governments, commercial and academic institutions as well as major sports teams.
Dysrupt Labs is based in Melbourne, Australia, with its research partners, The Florey Institute of Neuroscience & Mental Health and The University of Melbourne - Brain, Mind & Markets Laboratory.
Expert groups are making better forecasts with Almanis
Almanis is an AI-enhanced SaaS platform for improving the forecasts of expert groups by knowing who to listen to, and when.
Accuracy is superior to conventional group forecast mechanisms.
Speed improvement is significant. Your experts can be up to 15 days ahead of other forecast methods.
Expert groups unlock alpha and better manage uncertainty with Almanis
Almanis forecasts in:
corporate revenue & earnings
geopolitics
macroeconomics
public health, &
others
This is all in real-time and within the transparent control of your people, fully supported 24/7 by Almanis under an all-inclusive monthly subscription.
Almanis performance is verified
Almanis performance has been independently replicated on out-of-sample datasets from the DARPA-funded NG2S program.
Almanis advantage is unique
-
Where members have an equal say.
-
Where weighting favours past performance.
-
Where the group members self-weight conviction in real-time.
-
Combining 2 & 3 Almanis adds unique behavioural signatures that determine which forecasts are more likely to contain new, consensus-changing information [“wisdom in the crowd”].
Almanis heuristics are reliable
-
If Almanis hybrid agrees with an expert group, the expert group consensus is right.
-
The greater the disagreement between the expert group and the Almanis hybrid, the more likely the hybrid is to be right. This is particularly true in real-world times of high market uncertainty or volatility. Or where Taleb’s “fat-tails” are emerging..
-
If the Almanis hybrid and the expert group are on opposite sides of a forecast, i.e. the experts say heads and the hybrid says tails, the hybrid will be consistently more accurate.
3 ways expert groups are using Almanis
-
Make superior human-machine hybrid forecasts while asking more & better questions by bringing the Almanis platform in-house for your experts.
Augmenting your experts
Our software, in your firm
Easy, flexible, & scalable
-
Utilise our experts' human-machine hybrid forecasts with live-feed data access
Leverage our panel of proven forecasters
Run parallel to your experts
List questions privately or publicly on our platform
-
Stay up to date and better informed, with newsletter highlights of our expert's hybrid forecasts
Preview a selection of our top-performing forecasts
Covering geopolitics, macroeconomics, corporate revenue/earnings, & public health
Next steps to using Almanis alt-data or Almanis SaaS
Almanis is used by expert groups who are already making forecasts like US Non-Farm Payroll, Apple's Revenue, elections, and military conflicts.
In such cases, the effort and investment in using Almanis are at the margin and leverage institutional sunk cost.
The choice of next steps is yours:
-
To speed you on your way, let's have a quick chat so we can customise the materials for your time, effort and domain. We can see what will work in your decision processes - alt-data from our experts, or yours with our SaaS.
We will then send you a standard PDF with references and links to key resources including our newsletter on Substack.
From here, you can review our research and academic papers together with our contemporaneous record in Substack. Talk to us about your questions.
-
We can then talk with you about how you want to approach your investigation. We will provide you with the data set, data dictionary and sample testing protocols [in Stata & R] with related white-papers.
-
A turnkey experience from set-up to analysis and conclusions. An easy, safe experiment with 30 colleagues on a limited question set for free over 90 days - or a 1,000-user unlimited trial over 120 days.
The experiment or trial phase involves the basics of set-up. We work with you to easily settle your final formats and integrations for smooth scale-up. Service agreements are straightforward, and there are no lock-in contracts on your monthly subscription.
Overheads depend on the question format.
Regular recurring questions such as US Non Farm Payroll announcements, are low maintenance after initial set-up.
Irregular, highly granular one-offs can be high maintenance, and set-up costs can vary (so presumably high payoff for your effort).
Almanis SaaS can be operated one of two ways: we operate the platform for you - the forecasters, the questions, the training, the operation, everything. Or you take all or part of the operations as it suits your business.
Become a forecaster on Almanis
Challenge yourself, and earn financial rewards for performance.
At Almanis by Dysrupt Labs, we pride ourselves on the quality of our 800-plus forecasters from a diverse group of competitive participants globally.
This team have a dedication to dimensioning early and accurate insight.
The top forecasters can earn GBP 5 figures per annum in macroeconomics/geopolitics/public health and GBP 6 figures per annum in Corporate Events.
Free to forecast. Always.