13. Data Collection Wavedata founded in 2000 60,000 hours of data entry 160 wholesalers and suppliers Thousands of generic products 2 years of analysis
20. First Model (2005) 80 products analysed Linear dynamic model 3 forecast models A, B and C Based on statistical coefficients
21. Further development (2006) Another year of modelling 120 products analysed N on-linear polynomial model Adding Reimbursement arguments Including Tariff M
22. Current model completed (2007) Works for 99% of products No therapeutic adjustment needed No strength adjustment needed Integrated into a web site
38. Key Bounce factors Cost of goods Manufacturer withdrawal Short or long residual life Holiday link? Bounces are visible side of seasonality? Disease timings ie hay fever
39. How often bounces happen 282 products analysed 42 products bounced once 6 products bounced twice 4 products bounced three times 18% of products bounced
#3: Research project Trying to find the hidden patterns behind Long term Points interesting / surprising / unexpected
#4: How do companies set or predict prices. What would you do?
#5: A guess could be based on a particular competitor, or a key supplier, or an average, but it is still at heart a guess
#6: Trend lines can be added to what data you have, but often more than one line can be fitted to the same trend, and if you only have a few months of post patent generic prices, many trends could be fitted to the same data each one a guess.
#7: The price went up, as it was following a hidden pattern, not immediately obvious to the uninformed
#8: Trend lines can be added to what data you have, but often more than one line can be fitted to the same trend, and if you only have a few months of post patent generic prices, many trends could be fitted to the same data each one a guess.
#9: Trend lines can be added to what data you have, but often more than one line can be fitted to the same trend, and if you only have a few months of post patent generic prices, many trends could be fitted to the same data each one a guess.
#18: Each could be connected with the underlying factors
#19: Even the rises in price due to availability and price could be predicted
#20: Even the rises in price due to availability and price could be predicted
#21: The range of decline curves possible is very wide
#22: The range of decline curves possible is very wide
#24: Patterns are complex but predictable Not as complicated as shares But pharmaceuticals is an island doesn't pick up techniques from other industries easily Why has no one done it before not enough pricing history Norton healthcare SPSS experience
#32: So the method works Will get better as more examples are analysed Sugar traders / stocks and shares But bounces still an issue .. But what about other markets
#37: Nothing is new But pharma is still left with vulnerability in those companies that dont take this on board No reason why techniques wont work in other states